Update utils/suggestions.py

utils/suggestions.py

@@ -1,60 +1,31 @@
-from transformers import AutoModelForCausalLM, AutoTokenizer
-import torch
+import pandas as pd
+import os
 
-# Load ungated Phi-2 model
-model_name = "microsoft/phi-2"
-tokenizer = AutoTokenizer.from_pretrained(model_name)
-model = AutoModelForCausalLM.from_pretrained(
-    model_name,
-    torch_dtype=torch.float32
-)
-model.eval()
+BASE_DIR = os.path.join(os.path.dirname(__file__), "..", "csvs")
 
-def call_llm(prompt):
-    inputs = tokenizer(prompt, return_tensors="pt", truncation=True)
-    with torch.no_grad():
-        outputs = model.generate(
-            inputs["input_ids"],
-            max_new_tokens=250,
-            do_sample=True,
-            temperature=0.7,
-            top_k=50
-        )
-    return tokenizer.decode(outputs[0], skip_special_tokens=True)
+def load_csv(filename):
+    path = os.path.join(BASE_DIR, filename)
+    return pd.read_csv(path)
 
-def get_certification_suggestions(cv_text):
-    prompt = f"""You are a helpful career assistant. Analyze the following CV content and suggest relevant certifications to improve the candidate's chances of getting international jobs.
+def match_field(field, df):
+    field = field.strip().lower()
+    match = df[df['field'].str.lower().str.contains(field)]
+    if not match.empty:
+        return match['suggestion'].tolist()
+    return ["No specific suggestions found for this field."]
 
-CV:
-{cv_text}
+def get_certification_suggestions(field):
+    df = load_csv("certifications.csv")
+    return match_field(field, df)
 
-List the top 5 certifications with short descriptions.
-"""
-    response = call_llm(prompt)
-    return response.strip()
+def get_higher_education_suggestions(field):
+    df = load_csv("education.csv")
+    return match_field(field, df)
 
-def get_higher_education_suggestions(cv_text):
-    prompt = f"""You are a career coach. Based on the following CV, suggest 3 higher education degrees or programs that would significantly improve the candidate's job prospects internationally. Include degree names and reasons.
+def get_visa_recommendations(field):
+    df = load_csv("visa.csv")
+    return match_field(field, df)
 
-CV:
-{cv_text}
-"""
-    response = call_llm(prompt)
-    return response.strip()
-
-def get_visa_recommendations(cv_text):
-    prompt = f"""You are a global career visa advisor. Based on the following CV text, suggest 3 countries where the candidate has a high chance of qualifying for a work visa. Mention the type of visa, eligibility reasons, and links if possible.
-
-CV:
-{cv_text}
-"""
-    response = call_llm(prompt)
-    return response.strip()
-
-def get_career_advice(cv_text):
-    prompt = f"""You are a professional career counselor. Analyze the following CV text and give personalized career advice. Mention areas of improvement, potential career paths, and tips for international job hunting.
-
-CV:
-{cv_text}
-"""
-    return call_llm(prompt).strip()
+def get_career_advice(field):
+    df = load_csv("career_advice.csv")
+    return match_field(field, df)