AI Career Companion: Generative Guidance System

Introduction

Traditional career counseling is often expensive and inaccessible. Our objective was to engineer a Context-Aware Conversational Agent that could provide university-level guidance at zero marginal cost.

System Architecture

The system is built on a RAG (Retrieval-Augmented Generation) adjacent framework using Google's Gemini Pro.

LLM Engine: Gemini Pro for reasoning and roadmap generation.
State Management: Custom session handling to remember user constraints (GPA, interests, location) across the conversation.
Analytics: Logs user queries to CSV for identifying common student pain points.

Context Retention Logic

To prevent the "hallucination" of user details, we implemented a sliding window context buffer:

def update_history(user_input, ai_response):
    st.session_state.history.append({"role": "user", "content": user_input})
    st.session_state.history.append({"role": "model", "content": ai_response})
    
    # Export interaction log for analytics
    log_interaction(user_input, ai_response, timestamp=time.now())

Capabilities

Resume Scanning: Extracts keywords to suggest missing skills.
Roadmap Generation: Creates week-by-week study plans for specific roles (e.g., "DevOps Engineer").
Mock Interviews: Simulates role-specific interview questions.

Results

The system successfully handled diverse career queries, providing actionable roadmaps that aligned with current industry standards (verified against job descriptions).