Case Study
LinkedIn AI Copilot
A personal AI-powered LinkedIn assistant Chrome extension that generates smart comments, analyzes job postings, batch-scores jobs, and enhances your profile — powered by Groq + Llama 3. Utilizes Groq's high-speed inference engine for near-instant AI responses compared to standard APIs.
AI Copilot
The Problem
LinkedIn professionals spend hours crafting engaging comments, manually comparing job postings to their skills, and struggling to optimize their profiles for recruiters. The process is fragmented across multiple tools and extremely time-consuming.
My Approach
Chose FastAPI for high-performance async API serving. Integrated Groq with Llama 3 for fast, intelligent content generation with structured prompt engineering. Chrome Extension (Manifest V3) architecture enables real-time interaction directly on LinkedIn pages. SQLite for persistent profile storage.
Architecture
FastAPI backend serves RESTful endpoints for comment generation, job analysis, batch scoring, and profile enhancement. The Chrome Extension uses a service worker (background.js) as an API proxy while content scripts extract data directly from LinkedIn's DOM. Groq + Llama 3 powers all AI features with carefully engineered prompt templates. Profile data persists in an async SQLite database via pydantic-settings configuration. The Job Scanner module injects color-coded match-score badges directly onto LinkedIn job cards.
Challenges
Reliable DOM parsing across LinkedIn's dynamic, frequently-changing page structure was the core challenge. Built modular parser components (job-card-parser, job-page-detector) that can be updated independently. Prompt engineering for consistent, structured JSON output across diverse comment styles and job descriptions required iterative refinement. Implementing batch job scoring with real-time badge injection needed careful orchestration between content scripts and the background service worker.
Outcome
Reduced average job application evaluation time from 45 minutes to under 10 minutes. The batch Job Scanner feature lets users score an entire page of jobs at once with visual badges. Profile Enhancement provides structured, section-by-section optimization with a 0–10 scoring system and ranked improvement priorities.