import os import json import uvicorn from fastapi import FastAPI, Request from fastapi.responses import HTMLResponse, StreamingResponse from huggingface_hub import hf_hub_download from llama_cpp import Llama # --- ENGINE SETUP --- # Using the 1B model for maximum speed on Free Tier repo_id = "unsloth/Llama-3.2-1B-Instruct-GGUF" filename = "Llama-3.2-1B-Instruct-Q4_K_M.gguf" print("🚀 Mahoba Logic: Downloading/Loading Model...") model_path = hf_hub_download(repo_id=repo_id, filename=filename) llm = Llama(model_path=model_path, n_ctx=2048, n_threads=4, verbose=False) print("✅ Engine Ready!") # --- THE FRONTEND --- HTML_CONTENT = """ Mayank's AI Career Guider | The 1% Filter

🎯 The 1% Filter

Stop following the crowd. Build your roadmap for career domination.

Mission Progress: 0%

""" # --- BACKEND LOGIC --- app = FastAPI() @app.get("/") async def get_ui(): return HTMLResponse(content=HTML_CONTENT) @app.post("/guide") async def guide(request: Request): data = await request.json() user_name = data.get("name", "Student") interest = data.get("interest", "Career") query = data.get("followup", "") # Optimized prompt for Llama 3.2 1B prompt = f"<|begin_of_text|><|start_header_id|>system<|end_header_id|>\n\nYou are the 1% Filter AI by Mayank. Helping {user_name} with {interest}. Be short, brutal, and strategic. Focus on real skills, not degrees.<|eot_id|><|start_header_id|>user<|end_header_id|>\n\n{query}<|eot_id|><|start_header_id|>assistant<|end_header_id|>\n\n" def stream_generator(): stream = llm( prompt, max_tokens=150, stream=True, stop=["<|eot_id|>", "User:"] ) for output in stream: if "choices" in output: token = output["choices"][0].get("text", "") yield json.dumps({"token": token}) + "\n" return StreamingResponse( stream_generator(), media_type="application/x-ndjson", headers={"Cache-Control": "no-cache", "X-Content-Type-Options": "nosniff"} ) if __name__ == "__main__": uvicorn.run(app, host="0.0.0.0", port=7860)