🧠 Competitive Programming Analyzer

A full-stack AI-powered platform to track, analyze, and interact with competitive programming data from LeetCode, Codeforces, and CodeChef.

This project combines:

• 🔁 Backend automation for scheduled contest data ingestion
• 📊 Frontend dashboards for insightful visual analytics
• 🧠 A natural language query interface using LLMs and LangGraph

---

🚀 Key Features

🧩 Backend: Automated Data Pipeline

• Fetches and stores student participation data from LeetCode, Codeforces, and CodeChef.
• Runs a scheduling logic (at midnight) to check if a contest ended on that day.
• If yes, schedules a follow-up fetch at an 8-hour offset to allow for rating updates.
• Data includes contest names, rankings, questions attempted, ratings, etc.
• Designed using Prisma ORM and a PostgreSQL database (hosted on Neon).

The backend is structured around a normalized relational schema linking students, contests, and platform-specific participations, with indexing and uniqueness constraints for integrity and performance.

---

📊 Frontend: Visual Analytics Dashboard

Built with Next.js, React, TailwindCSS, and Chart.js, the UI offers:

• Executive Summary: Quick view of average participation, top performers, and contest frequency.
• Leaderboard: Sortable rankings based on performance across all contests.
• Analysis Page: Line charts for attendance trends and bar charts for performance over time.
• Filters: Customize views by department, batch, contest platform, and more.

The frontend is focused on educators and mentors, helping them monitor engagement, track growth, and support students better.

---

🤖 Modal: Natural Language Query Engine

Powered by LangGraph, Ollama (LLaMA 3.2), and Gradio, this module lets users query the database like:

“Who has the highest Codeforces rating?”
“How many students are from each department?”

How it works:

1. User asks a natural language question via a Gradio UI.
2. The question is parsed by an LLM into an SQL query using LangGraph state management.
3. The SQL query is executed on the live Neon PostgreSQL DB using psycopg2.
4. The result is translated back into a readable response and displayed in the UI.

---

⚙️ Tech Stack

Layer	Tech Used
Frontend	Next.js, React, TailwindCSS, Chart.js
Backend	Node.js, Prisma ORM, PostgreSQL
AI Interface	LangGraph, LLaMA 3.2 (Ollama), Gradio
Database	PostgreSQL
Tools	Node Schedule, dotenv, ts-node, Python

---

🗓️ Cron & Scheduling Logic

• Midnight cron job: Scans for contests that ended "today".
• Scheduled fetch: For each contest, schedules a fetch task with an 8-hour delay to ensure updated ratings.
• Built using node-cron and node-schedule.

---

🧪 Example Questions (Modal)

• "Who participated in the most contests?"
• "Average LeetCode rating of 3rd-year students?"
• "Top 5 students by Codeforces rating?"
• "Which department has the highest average rating?"

---