• 🎓 BE in Artificial Intelligence & Machine Learning @ Thapar Institute of Engineering and Technology — graduating May 2028
• 💼 Research Intern @ TIET-UQ Centre of Excellence in Data Science & AI (June 2026 – July 2026)
• 💼 Research Intern @ TIET × Imperial College London — SCALED project (Dec 2025 – Mar 2026)
• 🔭 Currently building: MediCore — a real-time clinical backend (Kafka, Redis, TimescaleDB, FastAPI)
• 🌱 Currently learning: DSA, Backend Engineering
• 🏆 Kaggle Notebooks Expert — Top 1.35% (832 / 62,335)
• 📍 Punjab, India

• Developed a 4-stage preprocessing pipeline for cleaning and standardizing human-demonstrated robotic trajectories
• Implemented and evaluated 6 trajectory-learning models using LOO cross-validation for accuracy, shape similarity, and motion smoothness
• Developed a 4-model uncertainty-weighted ensemble to combine complementary trajectory predictions and improve motion smoothness

• SCALED project: Diagnosed and resolved a config-level checkpoint resumption failure, a float-index bug, and tensor shape mismatches across subdomain, latent, and global dimensions in the reconstruction pipeline
• Compared qualitative outputs of regression vs diffusion surrogate models, documenting stability, geometry adherence, and stochasticity tradeoffs

Production-grade backend system for real-time patient vitals monitoring.

• Kafka-based vitals ingestion over WebSocket with asyncio consumer worker pool
• Redis alert engine with TTL-based priority queuing
• TimescaleDB hypertable for time-series vitals storage
• JWT-secured FastAPI gateway, fully containerised with docker-compose

Built at HackMol 7.0, NIT Jalandhar. Multi-source signal aggregation with real-time alerting.

• FinBERT sentiment across 6 data streams + hybrid anomaly detection (Z-score + Isolation Forest)
• Prophet price forecasting with auto-generated briefs via Llama 3
• Live Streamlit dashboard + real-time Telegram alerts

Hybrid Ridge + LightGBM pipeline over 95k+ half-hourly demand records.

• Fourier seasonality, lag, and rolling-window feature engineering
• 13% RMSE reduction vs baseline | R² 0.85