- 🎓 Third-year CS student at Punjab Technical University, graduating May 2027
- 💻 I build full-stack systems (MERN) and train ML models that solve real problems
- 🔧 Cut API response time by 15% at my internship via MongoDB query optimization
- 🌐 Built a real-time chat app, WebRTC P2P video streamer & distributed rate limiter from scratch
- 🤖 Trained a NanoGPT-style language model, fraud detection classifier & fake news detection pipeline
- 📬 Looking for SDE / ML internships — Summer/Fall 2026
Frontend
Backend
Database
ML / AI
Tools
Token Bucket + Sliding Window algorithms · Node.js · Redis
Built a production-grade rate limiting system from scratch supporting both Token Bucket and Sliding Window strategies. Handles concurrent request bursts with configurable thresholds per route.
Socket.io · Node.js · MongoDB · React
Full-stack real-time messaging app with room-based chat, typing indicators, and persistent message history. Optimized for concurrent users with low-latency event delivery.
WebRTC · Node.js · Signaling server
Peer-to-peer video calling platform with optimized signaling logic for unstable network conditions. Implements ICE candidate handling and STUN/TURN fallback.
Python · Dictionary encoding · RLE compression
Built a columnar storage engine from scratch with dictionary encoding and Run-Length Encoding (RLE) compression — mimicking core concepts behind analytical databases like ClickHouse.
PyTorch · Transformer architecture · Character-level tokenization
Implemented a GPT-style autoregressive language model from scratch including multi-head self-attention, positional encoding, and training loop with learning rate scheduling.
Python · scikit-learn · Class imbalance handling (SMOTE)
Trained a fraud classifier on imbalanced financial data using SMOTE oversampling and ensemble methods, achieving high recall on minority class (fraud cases).
Open to SDE & ML internship opportunities · Summer/Fall 2026 · amankumar005562@gmail.com