Initial commit for Space deployment

.gitattributes

@@ -0,0 +1,38 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tar filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text
+backend/models/movies/*.npy filter=lfs diff=lfs merge=lfs -text
+backend/models/**/*.pt filter=lfs diff=lfs merge=lfs -text
+data/*.csv filter=lfs diff=lfs merge=lfs -text

.gitignore

@@ -0,0 +1,207 @@
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[codz]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py.cover
+.hypothesis/
+.pytest_cache/
+cover/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+.pybuilder/
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pyenv
+#   For a library or package, you might want to ignore these files since the code is
+#   intended to run in multiple environments; otherwise, check them in:
+# .python-version
+
+# pipenv
+#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
+#   However, in case of collaboration, if having platform-specific dependencies or dependencies
+#   having no cross-platform support, pipenv may install dependencies that don't work, or not
+#   install all needed dependencies.
+#Pipfile.lock
+
+# UV
+#   Similar to Pipfile.lock, it is generally recommended to include uv.lock in version control.
+#   This is especially recommended for binary packages to ensure reproducibility, and is more
+#   commonly ignored for libraries.
+#uv.lock
+
+# poetry
+#   Similar to Pipfile.lock, it is generally recommended to include poetry.lock in version control.
+#   This is especially recommended for binary packages to ensure reproducibility, and is more
+#   commonly ignored for libraries.
+#   https://python-poetry.org/docs/basic-usage/#commit-your-poetrylock-file-to-version-control
+#poetry.lock
+#poetry.toml
+
+# pdm
+#   Similar to Pipfile.lock, it is generally recommended to include pdm.lock in version control.
+#   pdm recommends including project-wide configuration in pdm.toml, but excluding .pdm-python.
+#   https://pdm-project.org/en/latest/usage/project/#working-with-version-control
+#pdm.lock
+#pdm.toml
+.pdm-python
+.pdm-build/
+
+# pixi
+#   Similar to Pipfile.lock, it is generally recommended to include pixi.lock in version control.
+#pixi.lock
+#   Pixi creates a virtual environment in the .pixi directory, just like venv module creates one
+#   in the .venv directory. It is recommended not to include this directory in version control.
+.pixi
+
+# PEP 582; used by e.g. github.com/David-OConnor/pyflow and github.com/pdm-project/pdm
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule
+celerybeat.pid
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.envrc
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+# pytype static type analyzer
+.pytype/
+
+# Cython debug symbols
+cython_debug/
+
+# PyCharm
+#  JetBrains specific template is maintained in a separate JetBrains.gitignore that can
+#  be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore
+#  and can be added to the global gitignore or merged into this file.  For a more nuclear
+#  option (not recommended) you can uncomment the following to ignore the entire idea folder.
+#.idea/
+
+# Abstra
+# Abstra is an AI-powered process automation framework.
+# Ignore directories containing user credentials, local state, and settings.
+# Learn more at https://abstra.io/docs
+.abstra/
+
+# Visual Studio Code
+#  Visual Studio Code specific template is maintained in a separate VisualStudioCode.gitignore 
+#  that can be found at https://github.com/github/gitignore/blob/main/Global/VisualStudioCode.gitignore
+#  and can be added to the global gitignore or merged into this file. However, if you prefer, 
+#  you could uncomment the following to ignore the entire vscode folder
+# .vscode/
+
+# Ruff stuff:
+.ruff_cache/
+
+# PyPI configuration file
+.pypirc
+
+# Cursor
+#  Cursor is an AI-powered code editor. `.cursorignore` specifies files/directories to
+#  exclude from AI features like autocomplete and code analysis. Recommended for sensitive data
+#  refer to https://docs.cursor.com/context/ignore-files
+.cursorignore
+.cursorindexingignore
+
+# Marimo
+marimo/_static/
+marimo/_lsp/
+__marimo__/

Dockerfile

@@ -0,0 +1,32 @@
+#Frontend
+FROM node:20-alpine AS fe
+WORKDIR /fe
+
+COPY frontend/package*.json ./
+RUN npm ci
+
+COPY frontend/ ./
+RUN npm run build
+
+
+#Backend
+FROM python:3.11-slim
+WORKDIR /app
+
+ENV HF_HOME=/tmp/hf \
+    TRANSFORMERS_CACHE=/tmp/hf/transformers \
+    SENTENCE_TRANSFORMERS_HOME=/tmp/hf/sentence_transformers \
+    PYTHONDONTWRITEBYTECODE=1 \
+    PYTHONUNBUFFERED=1
+
+COPY backend/ /app/backend/
+
+COPY backend/requirements.txt /app/requirements.txt
+RUN pip install --no-cache-dir -r /app/requirements.txt
+
+RUN mkdir -p /app/backend/static
+COPY --from=fe /fe/dist /app/backend/static
+
+EXPOSE 7860
+
+CMD ["uvicorn", "backend.main:app", "--host", "0.0.0.0", "--port", "7860"]

LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2025 Nagharjun Mathi Mariappan
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

README.md

@@ -0,0 +1,5 @@
+---
+title: Recommendation Engine
+sdk: docker
+app_port: 7860
+---

backend/engines/cards_engine.py

@@ -0,0 +1,108 @@
+import os
+import numpy as np
+import pandas as pd
+import torch
+import torch.nn as nn
+
+DEVICE = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+
+
+class TwoTower(nn.Module):
+    def __init__(self, user_dim, item_dim, hidden=128):
+        super().__init__()
+        self.user = nn.Sequential(
+            nn.Linear(user_dim, hidden),
+            nn.ReLU(),
+            nn.Linear(hidden, hidden),
+            nn.ReLU(),
+        )
+        self.item = nn.Sequential(
+            nn.Linear(item_dim, hidden),
+            nn.ReLU(),
+            nn.Linear(hidden, hidden),
+            nn.ReLU(),
+        )
+        self.head = nn.Sequential(
+            nn.Linear(hidden * 2, hidden),
+            nn.ReLU(),
+            nn.Linear(hidden, 1),
+        )
+
+    def forward(self, u, x):
+        ue = self.user(u)
+        ie = self.item(x)
+        z = torch.cat([ue, ie], dim=-1)
+        return self.head(z).squeeze(-1)
+
+
+def _user_vector(req):
+    pref_map = {"cashback": 0, "travel": 1, "points": 2, "none": 3}
+    onehot = np.zeros(4, dtype=np.float32)
+    onehot[pref_map.get(req.get("rewards_pref", "none"), 3)] = 1.0
+
+    income_log = np.log1p(float(req["annual_income"])) / 12.0
+    score_norm = (float(req["credit_score"]) - 300.0) / 550.0
+
+    spend = np.array(
+        [
+            float(req["spend_groceries"]),
+            float(req["spend_dining"]),
+            float(req["spend_gas"]),
+            float(req["spend_travel"]),
+            float(req["spend_online"]),
+        ],
+        dtype=np.float32,
+    )
+    spend = spend / (spend.sum() + 1e-6)
+
+    x = np.concatenate(
+        [
+            np.array(
+                [
+                    score_norm,
+                    income_log,
+                    1.0 if req["carry_balance"] else 0.0,
+                    1.0 if req["travel_abroad"] else 0.0,
+                    1.0 if req["no_annual_fee"] else 0.0,
+                    1.0 if req["balance_transfer"] else 0.0,
+                ],
+                dtype=np.float32,
+            ),
+            onehot,
+            spend,
+        ]
+    )
+    return x.astype(np.float32)
+
+
+class CardsEngine:
+    def __init__(self, model_dir):
+        self.model_dir = model_dir
+
+        meta_path = os.path.join(self.model_dir, "cards_meta.csv")
+        feats_path = os.path.join(self.model_dir, "card_item_features.npy")
+        weights_path = os.path.join(self.model_dir, "cards_two_tower.pt")
+
+        self.meta = pd.read_csv(meta_path)
+        self.item_features = np.load(feats_path).astype(np.float32)
+
+        user_dim = 15
+        item_dim = self.item_features.shape[1]
+
+        self.model = TwoTower(user_dim=user_dim, item_dim=item_dim, hidden=128).to(DEVICE)
+        self.model.load_state_dict(torch.load(weights_path, map_location=DEVICE))
+        self.model.eval()
+
+    def recommend(self, user_req, top_k=10):
+        uvec = _user_vector(user_req)
+        U = np.repeat(uvec[None, :], self.item_features.shape[0], axis=0)
+
+        with torch.no_grad():
+            u = torch.from_numpy(U).to(DEVICE)
+            x = torch.from_numpy(self.item_features).to(DEVICE)
+            score = torch.sigmoid(self.model(u, x)).cpu().numpy()
+
+        idx = np.argsort(score)[::-1][:top_k]
+        out = self.meta.iloc[idx].copy()
+        out["score"] = score[idx]
+        return out

backend/engines/movies_engine.py

@@ -0,0 +1,115 @@
+import os
+import numpy as np
+import pandas as pd
+import joblib
+import torch
+import torch.nn as nn
+from sentence_transformers import SentenceTransformer
+
+DEVICE = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+
+
+class MovieRanker(nn.Module):
+    def __init__(self, dim=384, hidden=256):
+        super().__init__()
+        self.net = nn.Sequential(
+            nn.Linear(dim * 2, hidden),
+            nn.ReLU(),
+            nn.Linear(hidden, hidden),
+            nn.ReLU(),
+            nn.Linear(hidden, 1),
+        )
+
+    def forward(self, u, m):
+        x = torch.cat([u, m], dim=-1)
+        return self.net(x).squeeze(-1)
+
+
+class MoviesEngine:
+    def __init__(self, model_dir):
+        self.model_dir = model_dir
+
+        movies_path = os.path.join(model_dir, "movies_meta.csv")
+        idmap_path = os.path.join(model_dir, "movie_id2idx.pkl")
+        emb_path = os.path.join(model_dir, "movie_emb.npy")
+        model_path = os.path.join(model_dir, "movie_ranker.pt")
+
+        self.movies = pd.read_csv(movies_path)
+        self.movie_id2idx = joblib.load(idmap_path)
+        self.movie_emb = np.load(emb_path).astype(np.float32)
+
+        self.embedder = SentenceTransformer("all-MiniLM-L6-v2")
+
+        self.model = MovieRanker(dim=self.movie_emb.shape[1], hidden=256).to(DEVICE)
+        self.model.load_state_dict(torch.load(model_path, map_location=DEVICE))
+        self.model.eval()
+
+        self._title_lower = (
+            self.movies["title"]
+            .fillna("")
+            .astype(str)
+            .str.lower()
+            .values
+        )
+
+    def search(self, q: str, limit=10, min_sim=0.40):
+        q = (q or "").strip()
+        if not q:
+            return []
+
+        q_emb = self.embedder.encode([q], show_progress_bar=False)
+        q_emb = np.asarray(q_emb[0], dtype=np.float32)
+
+        eps = 1e-8
+        q_norm = np.linalg.norm(q_emb) + eps
+        m_norm = np.linalg.norm(self.movie_emb, axis=1) + eps
+
+        sims = (self.movie_emb @ q_emb) / (m_norm * q_norm)
+
+        idx = np.argsort(sims)[::-1]
+        out = []
+        for i in idx:
+            if len(out) >= limit:
+                break
+            if float(sims[i]) < float(min_sim):
+                break
+
+            r = self.movies.iloc[i]
+            out.append({
+                "movieId": int(r["movieId"]),
+                "title": str(r["title"]),
+                "genres": r.get("genres", ""),
+                "similarity": float(sims[i]),
+            })
+
+        return out
+
+
+    def _user_embedding(self, genres, liked_movie_ids):
+        vecs = []
+        for mid in (liked_movie_ids or []):
+            idx = self.movie_id2idx.get(int(mid))
+            if idx is not None:
+                vecs.append(self.movie_emb[idx])
+
+        if vecs:
+            return np.mean(np.stack(vecs, axis=0), axis=0).astype(np.float32)
+
+        g = [x.strip() for x in (genres or []) if x and str(x).strip()]
+        text = "User likes: " + ", ".join(g) if g else "User likes: popular movies"
+        u = self.embedder.encode([text], show_progress_bar=False)
+        return np.asarray(u[0], dtype=np.float32)
+
+    def recommend(self, genres=None, liked_movie_ids=None, top_k=10):
+        uvec = self._user_embedding(genres or [], liked_movie_ids or [])
+        U = np.repeat(uvec[None, :], self.movie_emb.shape[0], axis=0)
+
+        with torch.no_grad():
+            u = torch.from_numpy(U).to(DEVICE)
+            m = torch.from_numpy(self.movie_emb).to(DEVICE)
+            scores = torch.sigmoid(self.model(u, m)).cpu().numpy()
+
+        idx = np.argsort(scores)[::-1][:top_k]
+        out = self.movies.iloc[idx].copy()
+        out["score"] = scores[idx]
+        return out

backend/main.py

@@ -0,0 +1,128 @@
+import os
+
+from fastapi import FastAPI
+from fastapi.middleware.cors import CORSMiddleware
+from pydantic import BaseModel, Field
+from typing import List, Literal, Optional
+
+from engines.movies_engine import MoviesEngine
+from engines.cards_engine import CardsEngine
+
+
+app = FastAPI()
+
+app.add_middleware(
+    CORSMiddleware,
+    allow_origins=[
+        "http://localhost:5173",
+        "http://127.0.0.1:5173",
+    ],
+    allow_credentials=True,
+    allow_methods=["*"],
+    allow_headers=["*"],
+)
+
+BASE_DIR = os.path.dirname(os.path.abspath(__file__))
+MOVIE_DIR = os.path.join(BASE_DIR, "models", "movies")
+CARD_DIR = os.path.join(BASE_DIR, "models", "cards")
+
+movies_engine = MoviesEngine(MOVIE_DIR)
+cards_engine = CardsEngine(CARD_DIR)
+
+
+@app.get("/api/health")
+def health():
+    return {"status": "ok"}
+
+class MovieRecommendReq(BaseModel):
+    genres: List[str] = []
+    liked_movie_ids: List[int] = []
+    top_k: int = Field(10, ge=1, le=50)
+
+
+class MovieRec(BaseModel):
+    movieId: int
+    title: str
+    genres: Optional[str] = None
+    score: float
+
+
+@app.get("/api/movies/search")
+def search_movies(q: str):
+    return movies_engine.search(q, limit=10)
+
+
+@app.post("/api/movies/recommend", response_model=List[MovieRec])
+def recommend_movies(req: MovieRecommendReq):
+    df = movies_engine.recommend(
+        genres=req.genres,
+        liked_movie_ids=req.liked_movie_ids,
+        top_k=req.top_k,
+    )
+
+    out = []
+    for _, r in df.iterrows():
+        out.append(
+            MovieRec(
+                movieId=int(r["movieId"]),
+                title=str(r["title"]),
+                genres=str(r.get("genres", "")) if r.get("genres", None) is not None else None,
+                score=float(r["score"]),
+            )
+        )
+    return out
+
+class CardRecommendReq(BaseModel):
+    credit_score: int = Field(..., ge=300, le=850)
+    annual_income: float = Field(..., gt=0)
+
+    carry_balance: bool = False
+    travel_abroad: bool = False
+    no_annual_fee: bool = False
+    balance_transfer: bool = False
+
+    rewards_pref: Literal["cashback", "travel", "points", "none"] = "cashback"
+
+    spend_groceries: float = 300
+    spend_dining: float = 250
+    spend_gas: float = 120
+    spend_travel: float = 80
+    spend_online: float = 200
+
+    top_k: int = Field(10, ge=1, le=50)
+
+
+class CardRec(BaseModel):
+    institution: Optional[str] = None
+    product: str
+    website: Optional[str] = None
+    phone: Optional[str] = None
+    score: float
+
+
+@app.post("/api/cards/recommend", response_model=List[CardRec])
+def recommend_cards(req: CardRecommendReq):
+    df = cards_engine.recommend(req.model_dump(), top_k=req.top_k)
+
+    out = []
+    for _, r in df.iterrows():
+        institution = str(r.get("Institution Name", "")).strip()
+        website = str(r.get("Website for Consumer", "")).strip()
+        phone = str(r.get("Telephone Number for Consumers", "")).strip()
+
+        out.append(
+            CardRec(
+                institution=institution if institution else None,
+                product=str(r.get("Product Name", "")),
+                website=website if website else None,
+                phone=phone if phone else None,
+                score=float(r["score"]),
+            )
+        )
+    return out
+
+from fastapi.staticfiles import StaticFiles
+
+STATIC_DIR = os.path.join(os.path.dirname(__file__), "static")
+if os.path.isdir(STATIC_DIR):
+    app.mount("/", StaticFiles(directory=STATIC_DIR, html=True), name="static")

backend/models/cards/card_item_features.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e18814d3e7eca0c511d8a2358bb1f656e2e28f0c8583602a8030c60512e2e6a0
+size 883088

backend/models/cards/cards_two_tower.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aa23b1d5e1cecf4e831d2cc0aaf84d40f51e3245961a520c928fb459dc0a1eed
+size 477519

backend/models/cards/num_scaler.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:10fbbed9e767cdd63536c8d1d40bca5d8e6507dd08df07db8ae2d02b5dbd680b
+size 759

backend/models/movies/idx2movie_id.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:99f333078ad4768677e13c1cb969eebfd3ea5371bc35083f8911c7229b7d658b
+size 718619

backend/models/movies/idx2user_id.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ffb6b556abfce977b8ca9c249c40ca16f4a736e410dae4cfa11a74a4ed725a0d
+size 29483

backend/models/movies/movie_id2idx.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:51d80f13fdf0cc4e94e43add3f585a89f160e9ffc2e867429ca8c5b66b5e11dd
+size 718619

backend/models/movies/movie_ranker.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:74709c7b4f29088b6a0e469d524b57fe4872b28dd91a43b57cc9219033b863de
+size 1054801

backend/models/movies/user_emb_train.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6bcc521abdbca3b3ee25679d89f84f6aefa3eee38c186846d7920efa168ef08e
+size 7672448

backend/models/movies/user_id2idx.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e641ead7a92b3d60c1c29f3a0bff59cb17971100b3f00b2e05e3cc206243f396
+size 29483

backend/requirements.txt

@@ -0,0 +1,9 @@
+fastapi==0.115.0
+uvicorn[standard]==0.30.6
+pydantic==2.9.2
+numpy==1.26.4
+pandas==2.2.2
+scikit-learn==1.5.2
+joblib==1.4.2
+torch==2.4.1
+sentence-transformers==3.0.1

backend/train_cards.py

@@ -0,0 +1,393 @@
+import os
+import numpy as np
+import pandas as pd
+import joblib
+import torch
+import torch.nn as nn
+from torch.utils.data import Dataset, DataLoader
+from sentence_transformers import SentenceTransformer
+from sklearn.preprocessing import StandardScaler
+
+BACKEND_DIR = os.path.dirname(os.path.abspath(__file__))
+PROJECT_ROOT = os.path.dirname(BACKEND_DIR)
+DATA_DIR = os.path.join(PROJECT_ROOT, "data")
+OUT_DIR = os.path.join(BACKEND_DIR, "models", "cards")
+os.makedirs(OUT_DIR, exist_ok=True)
+
+DEVICE = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+
+def to_float(x):
+    if pd.isna(x):
+        return np.nan
+    s = str(x).strip()
+    if not s:
+        return np.nan
+    s_lower = s.lower()
+    if s_lower in {"na", "n/a", "none"}:
+        return np.nan
+    s = s.replace("%", "")
+    try:
+        return float(s)
+    except Exception:
+        return np.nan
+
+
+def pick_apr_for_score(row, credit_score):
+    if credit_score < 600:
+        key = "Purchase APR poor"
+    elif credit_score < 720:
+        key = "Purchase APR good"
+    else:
+        key = "Purchase APR great"
+
+    v = to_float(row.get(key, np.nan))
+    if np.isnan(v):
+        v = to_float(row.get("Purchase APR median", np.nan))
+    if np.isnan(v):
+        v = 24.0
+    return float(v)
+
+
+def annual_fee(row):
+    v = to_float(row.get("Annual Fee", np.nan))
+    if np.isnan(v):
+        v = 0.0
+    return float(v)
+
+
+def _find_col(row, candidates):
+    for c in candidates:
+        if c in row and not pd.isna(row[c]):
+            return c
+    return None
+
+
+def foreign_fee_pct(row):
+    col = _find_col(row, [
+        "Foreign Transaction Fee (%)",
+        "Foreign Transaction Fee (%) ",
+        "Foreign Transaction Fee (%)\ufeff",
+    ])
+    if col is None:
+        return 0.0
+    v = to_float(row.get(col, np.nan))
+    if np.isnan(v):
+        return 0.0
+    return float(v)
+
+
+def has_balance_transfer(row):
+    s = str(row.get("Balance Transfer Offered?", "")).strip().lower()
+    return 1.0 if s in {"yes", "y", "true", "1"} else 0.0
+
+
+def bt_length_months(row):
+    v = to_float(row.get("Median Length of Balance Transfer APR", np.nan))
+    if np.isnan(v):
+        v = 0.0
+    return float(v)
+
+
+def intro_apr_months(row):
+    v = to_float(row.get("Median Length of Introductory APR", np.nan))
+    if np.isnan(v):
+        v = 0.0
+    return float(v)
+
+
+def build_card_text(df):
+    inst = df.get("Institution Name", pd.Series([""] * len(df))).fillna("").astype(str)
+    name = df.get("Product Name", pd.Series([""] * len(df))).fillna("").astype(str)
+    rewards = df.get("Rewards", pd.Series([""] * len(df))).fillna("").astype(str)
+    features = df.get("Card Features", pd.Series([""] * len(df))).fillna("").astype(str)
+    services = df.get("Services", pd.Series([""] * len(df))).fillna("").astype(str)
+    return (inst + " | " + name + " | " + rewards + " | " + features + " | " + services).tolist()
+
+
+def sample_user(rng):
+    credit_score = int(rng.integers(520, 821))
+    income = float(
+        rng.choice(
+            [35000, 50000, 70000, 90000, 120000, 160000],
+            p=[0.12, 0.18, 0.24, 0.20, 0.16, 0.10],
+        )
+    )
+
+    carry_balance = float(rng.choice([0, 1], p=[0.55, 0.45]))
+    travel_abroad = float(rng.choice([0, 1], p=[0.75, 0.25]))
+    wants_no_fee = float(rng.choice([0, 1], p=[0.55, 0.45]))
+    wants_balance_transfer = float(rng.choice([0, 1], p=[0.65, 0.35]))
+
+    rewards_pref = rng.choice(["cashback", "travel", "points", "none"], p=[0.45, 0.20, 0.25, 0.10])
+
+    groceries = float(rng.integers(50, 600))
+    dining = float(rng.integers(50, 600))
+    gas = float(rng.integers(0, 300))
+    travel = float(rng.integers(0, 500))
+    online = float(rng.integers(50, 600))
+
+    return {
+        "credit_score": credit_score,
+        "income": income,
+        "carry_balance": carry_balance,
+        "travel_abroad": travel_abroad,
+        "wants_no_fee": wants_no_fee,
+        "wants_balance_transfer": wants_balance_transfer,
+        "rewards_pref": rewards_pref,
+        "groceries": groceries,
+        "dining": dining,
+        "gas": gas,
+        "travel": travel,
+        "online": online,
+    }
+
+
+def user_vector(u):
+    pref_map = {"cashback": 0, "travel": 1, "points": 2, "none": 3}
+    onehot = np.zeros(4, dtype=np.float32)
+    onehot[pref_map[u["rewards_pref"]]] = 1.0
+
+    income_log = np.log1p(u["income"]) / 12.0
+    score_norm = (u["credit_score"] - 300.0) / 550.0
+
+    spend = np.array([u["groceries"], u["dining"], u["gas"], u["travel"], u["online"]], dtype=np.float32)
+    spend = spend / (spend.sum() + 1e-6)
+
+    x = np.concatenate(
+        [
+            np.array(
+                [
+                    score_norm,
+                    income_log,
+                    u["carry_balance"],
+                    u["travel_abroad"],
+                    u["wants_no_fee"],
+                    u["wants_balance_transfer"],
+                ],
+                dtype=np.float32,
+            ),
+            onehot,
+            spend,
+        ]
+    )
+    return x.astype(np.float32)
+
+
+def teacher_utility(u, row):
+    apr = pick_apr_for_score(row, u["credit_score"])
+    af = annual_fee(row)
+    fx = foreign_fee_pct(row)
+    bt_ok = has_balance_transfer(row)
+    bt_len = bt_length_months(row)
+    intro_len = intro_apr_months(row)
+
+    text = (str(row.get("Rewards", "")) + " " + str(row.get("Card Features", ""))).lower()
+
+    rewards_boost = 0.0
+    if u["rewards_pref"] == "cashback" and "cash" in text:
+        rewards_boost += 0.8
+    if u["rewards_pref"] == "travel" and ("travel" in text or "air" in text or "hotel" in text):
+        rewards_boost += 0.8
+    if u["rewards_pref"] == "points" and ("points" in text or "miles" in text):
+        rewards_boost += 0.8
+
+    apr_pen = apr / 30.0 if u["carry_balance"] > 0.5 else apr / 80.0
+    fee_pen = af / 300.0 if u["wants_no_fee"] > 0.5 else af / 800.0
+    fx_pen = (fx / 5.0) if u["travel_abroad"] > 0.5 else (fx / 20.0)
+
+    bt_boost = 0.0
+    if u["wants_balance_transfer"] > 0.5:
+        bt_boost += 0.6 * bt_ok + 0.02 * bt_len + 0.015 * intro_len
+
+    spend_boost = 0.0
+    if u["groceries"] > 250 and "grocery" in text:
+        spend_boost += 0.2
+    if u["dining"] > 250 and ("dining" in text or "restaurant" in text):
+        spend_boost += 0.2
+    if u["travel"] > 200 and "travel" in text:
+        spend_boost += 0.2
+
+    return float(rewards_boost + spend_boost + bt_boost - (apr_pen + fee_pen + fx_pen))
+
+
+class TwoTower(nn.Module):
+    def __init__(self, user_dim, item_dim, hidden=128):
+        super().__init__()
+        self.user = nn.Sequential(
+            nn.Linear(user_dim, hidden),
+            nn.ReLU(),
+            nn.Linear(hidden, hidden),
+            nn.ReLU(),
+        )
+        self.item = nn.Sequential(
+            nn.Linear(item_dim, hidden),
+            nn.ReLU(),
+            nn.Linear(hidden, hidden),
+            nn.ReLU(),
+        )
+        self.head = nn.Sequential(
+            nn.Linear(hidden * 2, hidden),
+            nn.ReLU(),
+            nn.Linear(hidden, 1),
+        )
+
+    def forward(self, u, x):
+        ue = self.user(u)
+        ie = self.item(x)
+        z = torch.cat([ue, ie], dim=-1)
+        return self.head(z).squeeze(-1)
+
+
+class PairDataset(Dataset):
+    def __init__(self, U, X, y):
+        self.U = U
+        self.X = X
+        self.y = y
+
+    def __len__(self):
+        return len(self.y)
+
+    def __getitem__(self, i):
+        return (
+            torch.from_numpy(self.U[i]),
+            torch.from_numpy(self.X[i]),
+            torch.tensor(self.y[i], dtype=torch.float32),
+        )
+
+
+def main():
+    csv_path = os.path.join(DATA_DIR, "tccp_cards.csv")
+    df = pd.read_csv(csv_path, low_memory=False)
+
+    df = df[df["Product Name"].notna()].copy()
+    df = df.reset_index(drop=True)
+
+    embedder = SentenceTransformer("all-MiniLM-L6-v2")
+    text_list = build_card_text(df)
+    text_emb = embedder.encode(text_list, batch_size=64, show_progress_bar=True)
+    text_emb = np.asarray(text_emb, dtype=np.float32)
+
+    num = []
+    for _, row in df.iterrows():
+        apr_median = to_float(row.get("Purchase APR median", np.nan))
+        if np.isnan(apr_median):
+            apr_median = 24.0
+        num.append(
+            [
+                float(apr_median),
+                annual_fee(row),
+                foreign_fee_pct(row),
+                float(has_balance_transfer(row)),
+                bt_length_months(row),
+                intro_apr_months(row),
+            ]
+        )
+    num = np.asarray(num, dtype=np.float32)
+
+    scaler = StandardScaler()
+    num_scaled = scaler.fit_transform(num).astype(np.float32)
+
+    item_features = np.concatenate([text_emb, num_scaled], axis=1).astype(np.float32)
+
+    rng = np.random.default_rng(42)
+
+    users = [sample_user(rng) for _ in range(30000)]
+    U, X, y = [], [], []
+
+    n_items = len(df)
+
+    for u in users:
+        uvec = user_vector(u)
+
+        utils = np.empty(n_items, dtype=np.float32)
+        for i in range(n_items):
+            utils[i] = teacher_utility(u, df.iloc[i])
+
+        top_idx = utils.argsort()[::-1][:5]
+        neg_pool = utils.argsort()[:2000] if n_items >= 2000 else utils.argsort()[: max(1, n_items // 3)]
+        neg_idx = rng.choice(neg_pool, size=min(20, len(neg_pool)), replace=False)
+
+        for i in top_idx:
+            U.append(uvec)
+            X.append(item_features[i])
+            y.append(1.0)
+
+        for i in neg_idx:
+            U.append(uvec)
+            X.append(item_features[i])
+            y.append(0.0)
+
+    U = np.asarray(U, dtype=np.float32)
+    X = np.asarray(X, dtype=np.float32)
+    y = np.asarray(y, dtype=np.float32)
+
+    idx = np.arange(len(y))
+    rng.shuffle(idx)
+    split = int(len(y) * 0.9)
+    tr, va = idx[:split], idx[split:]
+
+    train_ds = PairDataset(U[tr], X[tr], y[tr])
+    val_ds = PairDataset(U[va], X[va], y[va])
+
+    train_loader = DataLoader(train_ds, batch_size=512, shuffle=True)
+    val_loader = DataLoader(val_ds, batch_size=1024, shuffle=False)
+
+    model = TwoTower(user_dim=U.shape[1], item_dim=X.shape[1], hidden=128).to(DEVICE)
+    opt = torch.optim.Adam(model.parameters(), lr=1e-3)
+    crit = nn.BCEWithLogitsLoss()
+
+    def run(loader, train=True):
+        model.train(train)
+        total, nobs = 0.0, 0
+        for u_b, x_b, y_b in loader:
+            u_b = u_b.to(DEVICE)
+            x_b = x_b.to(DEVICE)
+            y_b = y_b.to(DEVICE)
+
+            with torch.set_grad_enabled(train):
+                logit = model(u_b, x_b)
+                loss = crit(logit, y_b)
+                if train:
+                    opt.zero_grad()
+                    loss.backward()
+                    opt.step()
+
+            total += loss.item() * y_b.size(0)
+            nobs += y_b.size(0)
+
+        return total / max(1, nobs)
+
+    for epoch in range(5):
+        tr_loss = run(train_loader, True)
+        va_loss = run(val_loader, False)
+        print(f"epoch {epoch+1}  train_loss={tr_loss:.4f}  val_loss={va_loss:.4f}")
+
+    torch.save(model.state_dict(), os.path.join(OUT_DIR, "cards_two_tower.pt"))
+    np.save(os.path.join(OUT_DIR, "card_item_features.npy"), item_features)
+    joblib.dump(scaler, os.path.join(OUT_DIR, "num_scaler.pkl"))
+
+    meta_cols = [
+        "Institution Name",
+        "Product Name",
+        "Targeted Credit Tiers",
+        "Secured Card",
+        "Annual Fee",
+        "Purchase APR poor",
+        "Purchase APR good",
+        "Purchase APR great",
+        "Purchase APR median",
+        "Foreign Transaction Fee (%)",
+        "Website for Consumer",
+        "Telephone Number for Consumers",
+        "Rewards",
+        "Card Features",
+    ]
+    cols = [c for c in meta_cols if c in df.columns]
+    meta = df[cols].copy()
+    meta.to_csv(os.path.join(OUT_DIR, "cards_meta.csv"), index=False)
+
+    print("Saved cards model to:", OUT_DIR)
+
+
+if __name__ == "__main__":
+    main()

backend/train_movies.py

@@ -0,0 +1,166 @@
+import os
+import numpy as np
+import pandas as pd
+import joblib
+import torch
+import torch.nn as nn
+from torch.utils.data import Dataset, DataLoader
+from sentence_transformers import SentenceTransformer
+from sklearn.model_selection import train_test_split
+
+DEVICE = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+
+def build_movie_text(movies_df):
+    title = movies_df["title"].fillna("").astype(str)
+    genres = movies_df["genres"].fillna("").astype(str)
+    return (title + " | " + genres).tolist()
+
+
+def build_user_profile_embeddings(ratings_df, movie_id2idx, movie_emb, like_threshold=4.0):
+    liked = ratings_df[ratings_df["rating"] >= like_threshold]
+    grp = liked.groupby("userId")["movieId"].apply(list)
+
+    user_ids = grp.index.values
+    user_emb = np.zeros((len(user_ids), movie_emb.shape[1]), dtype=np.float32)
+
+    for i, uid in enumerate(user_ids):
+        mids = grp.loc[uid]
+        idxs = [movie_id2idx[m] for m in mids if m in movie_id2idx]
+        if idxs:
+            user_emb[i] = movie_emb[idxs].mean(axis=0)
+        else:
+            user_emb[i] = 0.0
+
+    user_id2idx = {int(uid): i for i, uid in enumerate(user_ids)}
+    return user_id2idx, user_emb
+
+
+class MoviePairDataset(Dataset):
+    def __init__(self, df, user_id2idx, movie_id2idx, user_emb, movie_emb):
+        df = df[df["userId"].isin(user_id2idx) & df["movieId"].isin(movie_id2idx)].copy()
+
+        self.u = df["userId"].map(user_id2idx).astype(int).values
+        self.m = df["movieId"].map(movie_id2idx).astype(int).values
+        self.y = (df["rating"].values >= 4.0).astype(np.float32)
+
+        self.user_emb = user_emb
+        self.movie_emb = movie_emb
+
+    def __len__(self):
+        return len(self.y)
+
+    def __getitem__(self, i):
+        uvec = self.user_emb[self.u[i]]
+        mvec = self.movie_emb[self.m[i]]
+        y = self.y[i]
+        return (
+            torch.from_numpy(uvec),
+            torch.from_numpy(mvec),
+            torch.tensor(y, dtype=torch.float32),
+        )
+
+
+class MovieRanker(nn.Module):
+    def __init__(self, dim=384, hidden=256):
+        super().__init__()
+        self.net = nn.Sequential(
+            nn.Linear(dim * 2, hidden),
+            nn.ReLU(),
+            nn.Linear(hidden, hidden),
+            nn.ReLU(),
+            nn.Linear(hidden, 1),
+        )
+
+    def forward(self, u, m):
+        x = torch.cat([u, m], dim=-1)
+        return self.net(x).squeeze(-1)
+
+
+def run_epoch(model, loader, optimizer, criterion, train=True):
+    model.train(train)
+    total_loss = 0.0
+    n = 0
+
+    for u, m, y in loader:
+        u, m, y = u.to(DEVICE), m.to(DEVICE), y.to(DEVICE)
+
+        with torch.set_grad_enabled(train):
+            logit = model(u, m)
+            loss = criterion(logit, y)
+
+            if train:
+                optimizer.zero_grad()
+                loss.backward()
+                optimizer.step()
+
+        total_loss += loss.item() * y.size(0)
+        n += y.size(0)
+
+    return total_loss / max(1, n)
+
+
+def main():
+    data_dir = "../data"
+    out_dir = "./models/movies"
+
+    os.makedirs(out_dir, exist_ok=True)
+
+    ratings = pd.read_csv(os.path.join(data_dir, "ratings.csv"))
+    movies = pd.read_csv(os.path.join(data_dir, "movies.csv"))
+
+    selected_users = sorted(ratings["userId"].unique())[:5000]
+    ratings = ratings[ratings["userId"].isin(selected_users)].copy()
+
+    embedder = SentenceTransformer("all-MiniLM-L6-v2")
+    movie_text = build_movie_text(movies)
+    movie_emb = embedder.encode(movie_text, batch_size=128, show_progress_bar=True)
+    movie_emb = np.asarray(movie_emb, dtype=np.float32)
+
+    movie_id2idx = {int(mid): i for i, mid in enumerate(movies["movieId"].values)}
+    idx2movie_id = {i: mid for mid, i in movie_id2idx.items()}
+
+    train_df, val_df = train_test_split(
+        ratings,
+        test_size=0.2,
+        random_state=42,
+        stratify=(ratings["rating"] >= 4.0).astype(int),
+    )
+
+    user_id2idx, user_emb = build_user_profile_embeddings(train_df, movie_id2idx, movie_emb)
+    idx2user_id = {i: uid for uid, i in user_id2idx.items()}
+
+    train_df = train_df[train_df["userId"].isin(user_id2idx) & train_df["movieId"].isin(movie_id2idx)].copy()
+    val_df = val_df[val_df["userId"].isin(user_id2idx) & val_df["movieId"].isin(movie_id2idx)].copy()
+
+    train_ds = MoviePairDataset(train_df, user_id2idx, movie_id2idx, user_emb, movie_emb)
+    val_ds = MoviePairDataset(val_df, user_id2idx, movie_id2idx, user_emb, movie_emb)
+
+    train_loader = DataLoader(train_ds, batch_size=512, shuffle=True)
+    val_loader = DataLoader(val_ds, batch_size=1024, shuffle=False)
+
+    model = MovieRanker(dim=movie_emb.shape[1], hidden=256).to(DEVICE)
+    optimizer = torch.optim.Adam(model.parameters(), lr=1e-3)
+    criterion = nn.BCEWithLogitsLoss()
+
+    for epoch in range(6):
+        tr = run_epoch(model, train_loader, optimizer, criterion, train=True)
+        va = run_epoch(model, val_loader, optimizer, criterion, train=False)
+        print(f"epoch {epoch+1}  train_loss={tr:.4f}  val_loss={va:.4f}")
+
+    torch.save(model.state_dict(), os.path.join(out_dir, "movie_ranker.pt"))
+    np.save(os.path.join(out_dir, "movie_emb.npy"), movie_emb)
+    np.save(os.path.join(out_dir, "user_emb_train.npy"), user_emb)
+
+    movies_small = movies[["movieId", "title", "genres"]].copy()
+    movies_small.to_csv(os.path.join(out_dir, "movies_meta.csv"), index=False)
+
+    joblib.dump(movie_id2idx, os.path.join(out_dir, "movie_id2idx.pkl"))
+    joblib.dump(idx2movie_id, os.path.join(out_dir, "idx2movie_id.pkl"))
+    joblib.dump(user_id2idx, os.path.join(out_dir, "user_id2idx.pkl"))
+    joblib.dump(idx2user_id, os.path.join(out_dir, "idx2user_id.pkl"))
+
+    print("Saved movies model to:", out_dir)
+
+
+if __name__ == "__main__":
+    main()

data/movies.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b37ca1abc7798de741138ed252b62f69f7e37c84b8a8fab1b82d409b4c6c5cc2
+size 4242926

data/tccp_cards.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c81ac7a3b3386cfecff1645b89f02a55360a85892901e4b4893cbb95c5747b74
+size 915779

frontend/.gitignore

@@ -0,0 +1,24 @@
+# Logs
+logs
+*.log
+npm-debug.log*
+yarn-debug.log*
+yarn-error.log*
+pnpm-debug.log*
+lerna-debug.log*
+
+node_modules
+dist
+dist-ssr
+*.local
+
+# Editor directories and files
+.vscode/*
+!.vscode/extensions.json
+.idea
+.DS_Store
+*.suo
+*.ntvs*
+*.njsproj
+*.sln
+*.sw?

frontend/README.md

@@ -0,0 +1,16 @@
+# React + Vite
+
+This template provides a minimal setup to get React working in Vite with HMR and some ESLint rules.
+
+Currently, two official plugins are available:
+
+- [@vitejs/plugin-react](https://github.com/vitejs/vite-plugin-react/blob/main/packages/plugin-react) uses [Babel](https://babeljs.io/) (or [oxc](https://oxc.rs) when used in [rolldown-vite](https://vite.dev/guide/rolldown)) for Fast Refresh
+- [@vitejs/plugin-react-swc](https://github.com/vitejs/vite-plugin-react/blob/main/packages/plugin-react-swc) uses [SWC](https://swc.rs/) for Fast Refresh
+
+## React Compiler
+
+The React Compiler is not enabled on this template because of its impact on dev & build performances. To add it, see [this documentation](https://react.dev/learn/react-compiler/installation).
+
+## Expanding the ESLint configuration
+
+If you are developing a production application, we recommend using TypeScript with type-aware lint rules enabled. Check out the [TS template](https://github.com/vitejs/vite/tree/main/packages/create-vite/template-react-ts) for information on how to integrate TypeScript and [`typescript-eslint`](https://typescript-eslint.io) in your project.

frontend/eslint.config.js

@@ -0,0 +1,29 @@
+import js from '@eslint/js'
+import globals from 'globals'
+import reactHooks from 'eslint-plugin-react-hooks'
+import reactRefresh from 'eslint-plugin-react-refresh'
+import { defineConfig, globalIgnores } from 'eslint/config'
+
+export default defineConfig([
+  globalIgnores(['dist']),
+  {
+    files: ['**/*.{js,jsx}'],
+    extends: [
+      js.configs.recommended,
+      reactHooks.configs.flat.recommended,
+      reactRefresh.configs.vite,
+    ],
+    languageOptions: {
+      ecmaVersion: 2020,
+      globals: globals.browser,
+      parserOptions: {
+        ecmaVersion: 'latest',
+        ecmaFeatures: { jsx: true },
+        sourceType: 'module',
+      },
+    },
+    rules: {
+      'no-unused-vars': ['error', { varsIgnorePattern: '^[A-Z_]' }],
+    },
+  },
+])

frontend/index.html

@@ -0,0 +1,13 @@
+<!doctype html>
+<html lang="en">
+  <head>
+    <meta charset="UTF-8" />
+    <link rel="icon" type="image/svg+xml" href="/vite.svg" />
+    <meta name="viewport" content="width=device-width, initial-scale=1.0" />
+    <title>frontend</title>
+  </head>
+  <body>
+    <div id="root"></div>
+    <script type="module" src="/src/main.jsx"></script>
+  </body>
+</html>

frontend/package.json

@@ -0,0 +1,27 @@
+{
+  "name": "frontend",
+  "private": true,
+  "version": "0.0.0",
+  "type": "module",
+  "scripts": {
+    "dev": "vite",
+    "build": "vite build",
+    "lint": "eslint .",
+    "preview": "vite preview"
+  },
+  "dependencies": {
+    "react": "^19.2.0",
+    "react-dom": "^19.2.0"
+  },
+  "devDependencies": {
+    "@eslint/js": "^9.39.1",
+    "@types/react": "^19.2.5",
+    "@types/react-dom": "^19.2.3",
+    "@vitejs/plugin-react": "^5.1.1",
+    "eslint": "^9.39.1",
+    "eslint-plugin-react-hooks": "^7.0.1",
+    "eslint-plugin-react-refresh": "^0.4.24",
+    "globals": "^16.5.0",
+    "vite": "^7.2.4"
+  }
+}

frontend/src/App.css

@@ -0,0 +1,42 @@
+#root {
+  max-width: 1280px;
+  margin: 0 auto;
+  padding: 2rem;
+  text-align: center;
+}
+
+.logo {
+  height: 6em;
+  padding: 1.5em;
+  will-change: filter;
+  transition: filter 300ms;
+}
+.logo:hover {
+  filter: drop-shadow(0 0 2em #646cffaa);
+}
+.logo.react:hover {
+  filter: drop-shadow(0 0 2em #61dafbaa);
+}
+
+@keyframes logo-spin {
+  from {
+    transform: rotate(0deg);
+  }
+  to {
+    transform: rotate(360deg);
+  }
+}
+
+@media (prefers-reduced-motion: no-preference) {
+  a:nth-of-type(2) .logo {
+    animation: logo-spin infinite 20s linear;
+  }
+}
+
+.card {
+  padding: 2em;
+}
+
+.read-the-docs {
+  color: #888;
+}

frontend/src/App.jsx

@@ -0,0 +1,15 @@
+import { useState } from "react";
+import NavBar from "./components/NavBar";
+import MoviesPage from "./pages/MoviesPage";
+import CardsPage from "./pages/CardsPage";
+import "./styles.css";
+
+export default function App() {
+  const [tab, setTab] = useState("movies");
+  return (
+    <div className="container">
+      <NavBar tab={tab} setTab={setTab} />
+      {tab === "movies" ? <MoviesPage /> : <CardsPage />}
+    </div>
+  );
+}

frontend/src/components/CardForm.jsx

@@ -0,0 +1,109 @@
+import { useState } from "react";
+
+export default function CardForm({ onSubmit }) {
+  const [v, setV] = useState({
+    credit_score: 720,
+    annual_income: 70000,
+    carry_balance: false,
+    travel_abroad: false,
+    no_annual_fee: true,
+    balance_transfer: false,
+    rewards_pref: "cashback",
+    spend_groceries: 300,
+    spend_dining: 250,
+    spend_gas: 120,
+    spend_travel: 80,
+    spend_online: 200,
+    top_k: 10,
+  });
+
+  function set(name, value) {
+    setV((p) => ({ ...p, [name]: value }));
+  }
+
+  return (
+    <form className="form" onSubmit={(e) => { e.preventDefault(); onSubmit(v); }}>
+      <div className="cardHeader">
+        <h2>Credit Cards</h2>
+      </div>
+
+      <div className="grid2">
+        <div className="field">
+          <label>Credit score</label>
+          <input
+            type="range"
+            min="300"
+            max="850"
+            value={v.credit_score}
+            onChange={(e) => set("credit_score", Number(e.target.value))}
+          />
+          <div className="muted small">{v.credit_score}</div>
+        </div>
+
+        <div className="field">
+          <label>Annual income</label>
+          <input
+            type="number"
+            value={v.annual_income}
+            onChange={(e) => set("annual_income", Number(e.target.value))}
+          />
+        </div>
+
+        <div className="field">
+          <label>Rewards</label>
+          <select value={v.rewards_pref} onChange={(e) => set("rewards_pref", e.target.value)}>
+            <option value="cashback">Cashback</option>
+            <option value="travel">Travel</option>
+            <option value="points">Points</option>
+            <option value="none">No preference</option>
+          </select>
+        </div>
+
+        <div className="field">
+          <label>Top results</label>
+          <input type="number" value={v.top_k} min="1" max="50" onChange={(e) => set("top_k", Number(e.target.value))} />
+        </div>
+      </div>
+
+      <div className="grid2">
+        <div className="field">
+          <label><input type="checkbox" checked={v.carry_balance} onChange={(e) => set("carry_balance", e.target.checked)} /> I carry a balance</label>
+        </div>
+        <div className="field">
+          <label><input type="checkbox" checked={v.travel_abroad} onChange={(e) => set("travel_abroad", e.target.checked)} /> I travel abroad</label>
+        </div>
+        <div className="field">
+          <label><input type="checkbox" checked={v.no_annual_fee} onChange={(e) => set("no_annual_fee", e.target.checked)} /> Prefer no annual fee</label>
+        </div>
+        <div className="field">
+          <label><input type="checkbox" checked={v.balance_transfer} onChange={(e) => set("balance_transfer", e.target.checked)} /> Need balance transfer</label>
+        </div>
+      </div>
+
+      <div className="grid2">
+        <div className="field">
+          <label>Groceries ($/mo)</label>
+          <input type="number" value={v.spend_groceries} onChange={(e) => set("spend_groceries", Number(e.target.value))} />
+        </div>
+        <div className="field">
+          <label>Dining ($/mo)</label>
+          <input type="number" value={v.spend_dining} onChange={(e) => set("spend_dining", Number(e.target.value))} />
+        </div>
+        <div className="field">
+          <label>Gas ($/mo)</label>
+          <input type="number" value={v.spend_gas} onChange={(e) => set("spend_gas", Number(e.target.value))} />
+        </div>
+        <div className="field">
+          <label>Travel ($/mo)</label>
+          <input type="number" value={v.spend_travel} onChange={(e) => set("spend_travel", Number(e.target.value))} />
+        </div>
+        <div className="field">
+          <label>Online ($/mo)</label>
+          <input type="number" value={v.spend_online} onChange={(e) => set("spend_online", Number(e.target.value))} />
+        </div>
+      </div>
+
+      <button className="primary" type="submit">Show cards</button>
+    </form>
+  );
+}

frontend/src/components/MovieForm.jsx

@@ -0,0 +1,75 @@
+import { useMemo, useState } from "react";
+
+const GENRES = [
+  "Action","Adventure","Animation","Comedy","Crime","Documentary","Drama","Fantasy",
+  "Horror","Mystery","Romance","Sci-Fi","Thriller"
+];
+
+export default function MovieForm({ onSubmit, onPickLikedMovie }) {
+  const [selected, setSelected] = useState([]);
+  const [search, setSearch] = useState("");
+
+  const hint = useMemo(() => {
+    if (!selected.length) return "Pick up to 3 genres.";
+    return `Genres: ${selected.join(", ")}`;
+  }, [selected]);
+
+  function toggle(g) {
+    setSelected((prev) => {
+      const has = prev.includes(g);
+      if (has) return prev.filter(x => x !== g);
+      return [...prev, g].slice(0, 3);
+    });
+  }
+
+  return (
+    <form
+      className="form"
+      onSubmit={(e) => {
+        e.preventDefault();
+        onSubmit({ genres: selected });
+      }}
+    >
+      <div className="cardHeader">
+        <h2>Movies</h2>
+        <p className="muted">{hint}</p>
+      </div>
+
+      <div className="field">
+        <label>Optional: add a movie you already like</label>
+        <input
+          value={search}
+          onChange={(e) => setSearch(e.target.value)}
+          placeholder='Search title (e.g. "Batman")'
+          onKeyDown={(e) => {
+            if (e.key === "Enter") {
+              e.preventDefault();
+              onPickLikedMovie(search);
+            }
+          }}
+        />
+        <button className="primary" type="button" onClick={() => onPickLikedMovie(search)}>
+          Search & add
+        </button>
+      </div>
+
+      <div className="field">
+        <label>Genres</label>
+        <div className="chips">
+          {GENRES.map((g) => (
+            <button
+              key={g}
+              type="button"
+              className={selected.includes(g) ? "chip on" : "chip"}
+              onClick={() => toggle(g)}
+            >
+              {g}
+            </button>
+          ))}
+        </div>
+      </div>
+
+      <button className="primary" type="submit">Get recommendations</button>
+    </form>
+  );
+}

frontend/src/components/NavBar.jsx

@@ -0,0 +1,19 @@
+export default function NavBar({ tab, setTab }) {
+  return (
+    <header className="topbar">
+      <div>
+        <div className="title">Recommendation Engine</div>
+        <div className="subtitle">Movies and Credit Cards</div>
+      </div>
+
+      <nav className="tabs">
+        <button className={tab === "movies" ? "tab active" : "tab"} onClick={() => setTab("movies")}>
+          Movies
+        </button>
+        <button className={tab === "cards" ? "tab active" : "tab"} onClick={() => setTab("cards")}>
+          Credit Cards
+        </button>
+      </nav>
+    </header>
+  );
+}

frontend/src/components/ResultsGrid.jsx

@@ -0,0 +1,16 @@
+export default function ResultsGrid({ items, renderTitle, renderMeta }) {
+  if (!items?.length) return null;
+
+  return (
+    <div className="results">
+      {items.map((it, idx) => (
+        <div className="item" key={idx}>
+          <div className="itemTitle">{renderTitle(it)}</div>
+          <div className="muted small" style={{ marginTop: 6 }}>
+            {renderMeta(it)}
+          </div>
+        </div>
+      ))}
+    </div>
+  );
+}

frontend/src/index.css

@@ -0,0 +1,68 @@
+:root {
+  font-family: system-ui, Avenir, Helvetica, Arial, sans-serif;
+  line-height: 1.5;
+  font-weight: 400;
+
+  color-scheme: light dark;
+  color: rgba(255, 255, 255, 0.87);
+  background-color: #242424;
+
+  font-synthesis: none;
+  text-rendering: optimizeLegibility;
+  -webkit-font-smoothing: antialiased;
+  -moz-osx-font-smoothing: grayscale;
+}
+
+a {
+  font-weight: 500;
+  color: #646cff;
+  text-decoration: inherit;
+}
+a:hover {
+  color: #535bf2;
+}
+
+body {
+  margin: 0;
+  display: flex;
+  place-items: center;
+  min-width: 320px;
+  min-height: 100vh;
+}
+
+h1 {
+  font-size: 3.2em;
+  line-height: 1.1;
+}
+
+button {
+  border-radius: 8px;
+  border: 1px solid transparent;
+  padding: 0.6em 1.2em;
+  font-size: 1em;
+  font-weight: 500;
+  font-family: inherit;
+  background-color: #1a1a1a;
+  cursor: pointer;
+  transition: border-color 0.25s;
+}
+button:hover {
+  border-color: #646cff;
+}
+button:focus,
+button:focus-visible {
+  outline: 4px auto -webkit-focus-ring-color;
+}
+
+@media (prefers-color-scheme: light) {
+  :root {
+    color: #213547;
+    background-color: #ffffff;
+  }
+  a:hover {
+    color: #747bff;
+  }
+  button {
+    background-color: #f9f9f9;
+  }
+}

frontend/src/main.jsx

@@ -0,0 +1,10 @@
+import { StrictMode } from 'react'
+import { createRoot } from 'react-dom/client'
+import './index.css'
+import App from './App.jsx'
+
+createRoot(document.getElementById('root')).render(
+  <StrictMode>
+    <App />
+  </StrictMode>,
+)

frontend/src/pages/CardsPage.jsx

@@ -0,0 +1,43 @@
+import { useState } from "react";
+import CardForm from "../components/CardForm";
+import ResultsGrid from "../components/ResultsGrid";
+
+const API = "";
+
+export default function CardsPage() {
+  const [recs, setRecs] = useState([]);
+  const [error, setError] = useState("");
+
+  async function submit(payload) {
+    setError("");
+    setRecs([]);
+
+    try {
+      const res = await fetch(`${API}/api/cards/recommend`, {
+        method: "POST",
+        headers: { "Content-Type": "application/json" },
+        body: JSON.stringify(payload),
+      });
+
+      if (!res.ok) throw new Error("Request failed");
+      setRecs(await res.json());
+    } catch (e) {
+      setError(e.message || "Error");
+    }
+  }
+
+  return (
+    <section className="card">
+      <CardForm onSubmit={submit} />
+      {error && <div className="muted small" style={{ marginTop: 12 }}>{error}</div>}
+
+      <ResultsGrid
+        items={recs}
+        renderTitle={(r) => r.product}
+        renderMeta={(r) =>
+          `${r.institution || "Institution"} • Score: ${r.score.toFixed(3)}`
+        }
+      />
+    </section>
+  );
+}

frontend/src/pages/MoviesPage.jsx

@@ -0,0 +1,70 @@
+import { useState } from "react";
+import MovieForm from "../components/MovieForm";
+import ResultsGrid from "../components/ResultsGrid";
+
+const API = "";
+
+export default function MoviesPage() {
+  const [liked, setLiked] = useState([]);
+  const [recs, setRecs] = useState([]);
+  const [msg, setMsg] = useState("");
+
+  async function addLikedMovie(q) {
+    const query = (q || "").trim();
+    if (!query) return;
+
+    const res = await fetch(`${API}/api/movies/search?q=${encodeURIComponent(query)}`);
+    const hits = await res.json();
+    if (!hits.length) {
+      setMsg("No matches found.");
+      return;
+    }
+
+    const first = hits[0];
+    setLiked((prev) => (prev.some(x => x.movieId === first.movieId) ? prev : [...prev, first]));
+    setMsg(`Added: ${first.title}`);
+  }
+
+  async function submit({ genres }) {
+    setMsg("");
+    const body = {
+      genres,
+      liked_movie_ids: liked.map(x => x.movieId),
+      top_k: 10,
+    };
+
+    const res = await fetch(`${API}/api/movies/recommend`, {
+      method: "POST",
+      headers: { "Content-Type": "application/json" },
+      body: JSON.stringify(body),
+    });
+
+    const data = await res.json();
+    setRecs(data);
+  }
+
+  return (
+    <section className="card">
+      <MovieForm onSubmit={submit} onPickLikedMovie={addLikedMovie} />
+
+      {liked.length > 0 && (
+        <div style={{ marginTop: 12 }}>
+          <div className="muted small">Liked movies:</div>
+          <div className="row" style={{ marginTop: 8 }}>
+            {liked.map((m) => (
+              <span className="badge" key={m.movieId}>{m.title}</span>
+            ))}
+          </div>
+        </div>
+      )}
+
+      {msg && <div className="muted small" style={{ marginTop: 12 }}>{msg}</div>}
+
+      <ResultsGrid
+        items={recs}
+        renderTitle={(r) => r.title}
+        renderMeta={(r) => `Score: ${r.score.toFixed(3)} • ${r.genres || "-"}`}
+      />
+    </section>
+  );
+}

frontend/src/styles.css

@@ -0,0 +1,85 @@
+:root {
+  --bg: #0b0c10;
+  --card: #12141b;
+  --text: #e9edf1;
+  --muted: #9aa3ad;
+  --line: #232733;
+  --accent: #6ea8ff;
+}
+
+* { box-sizing: border-box; }
+body {
+  margin: 0;
+  font-family: ui-sans-serif, system-ui, -apple-system, Segoe UI, Roboto, Arial;
+  background: var(--bg);
+  color: var(--text);
+}
+
+.container { max-width: 980px; margin: 0 auto; padding: 24px; }
+.topbar { display: flex; justify-content: space-between; align-items: center; gap: 16px; margin-bottom: 16px; }
+.title { font-size: 20px; font-weight: 650; }
+.subtitle { color: var(--muted); font-size: 13px; margin-top: 2px; }
+
+.tabs { display: flex; gap: 8px; }
+.tab {
+  background: transparent;
+  border: 1px solid var(--line);
+  color: var(--text);
+  padding: 8px 12px;
+  border-radius: 10px;
+  cursor: pointer;
+}
+.tab.active { border-color: var(--accent); }
+
+.card { background: var(--card); border: 1px solid var(--line); border-radius: 16px; padding: 18px; }
+.cardHeader h2 { margin: 0 0 4px 0; }
+.muted { color: var(--muted); }
+.small { font-size: 13px; }
+
+.form { margin-top: 14px; display: grid; gap: 12px; }
+.grid2 { display: grid; grid-template-columns: repeat(2, minmax(0, 1fr)); gap: 12px; }
+.field { display: grid; gap: 6px; }
+label { font-size: 13px; color: var(--muted); }
+
+input, select {
+  background: #0e1016;
+  border: 1px solid var(--line);
+  color: var(--text);
+  padding: 10px 12px;
+  border-radius: 12px;
+  outline: none;
+}
+input:focus, select:focus { border-color: var(--accent); }
+
+.primary {
+  background: var(--accent);
+  color: #0b0c10;
+  border: 0;
+  padding: 10px 14px;
+  border-radius: 12px;
+  cursor: pointer;
+  font-weight: 600;
+}
+
+.chips { display: flex; flex-wrap: wrap; gap: 8px; }
+.chip {
+  border: 1px solid var(--line);
+  background: transparent;
+  color: var(--text);
+  padding: 6px 10px;
+  border-radius: 999px;
+  cursor: pointer;
+  font-size: 13px;
+}
+.chip.on { border-color: var(--accent); }
+
+.results { margin-top: 16px; display: grid; grid-template-columns: repeat(2, minmax(0, 1fr)); gap: 12px; }
+.item { border: 1px solid var(--line); border-radius: 14px; padding: 12px; background: #0e1016; }
+.itemTitle { font-weight: 650; }
+.row { display: flex; gap: 8px; align-items: center; flex-wrap: wrap; }
+.badge { font-size: 12px; color: var(--muted); border: 1px solid var(--line); padding: 4px 8px; border-radius: 999px; }
+
+@media (max-width: 720px) {
+  .grid2 { grid-template-columns: 1fr; }
+  .results { grid-template-columns: 1fr; }
+}

frontend/vite.config.js

@@ -0,0 +1,7 @@
+import { defineConfig } from 'vite'
+import react from '@vitejs/plugin-react'
+
+// https://vite.dev/config/
+export default defineConfig({
+  plugins: [react()],
+})