About
Turning high-stakes problems into production-grade AI systems.
I’m Kavin Thakur, a Computer Science (AIML) undergraduate at Chitkara University (Minor in AI, IIT Ropar). As Tech Head — IEEE CUIET, a 7× Hackathon Champion, Harvard Aspire Fellow, and former Data Science Intern @ Infomo, I design and ship scalable AI, Web3, CV, and FinTech systems that perform under pressure.
Core work
VocalWell — AI voice-disorder detection (Librosa, TensorFlow/PyTorch). 90%+ accuracy, spectrogram & MFCC analysis, interactive clinical reports for early, non-invasive screening.
Agentic RAG Pipeline — Production RAG: hybrid FAISS + Redis retrieval, parallel LLM calls, semantic chunking, knowledge-graph integration, async processing, caching, cloud deployment, monitoring.
Player Re-Identification — YOLOv11 + IoU tracking + HSV-histogram ReID for persistent player IDs under occlusion; modular pipeline for automated sports analytics.
IEEE Event Systems — Live leaderboards, secure multi-role logins, projector-safe dashboards, and live game ops; handled 300+ live users and 10k+ simulated load with zero downtime.
Other flagship projects IEEE CUIET website + RAG chatbot; Mission Launch: Burn or Glory (Next.js, Hardhat, ERC-721/ERC-20); Fusion Pay (Web2.5 cross-border payments); AlphaMinds (AI education); PassKeys (Django password manager); Arvya (women’s fintech + AI chatbot); FLIX (Flask movie app); OWL (decentralized edtech).
Tech: Python, FastAPI, Flask, Django, Next.js, Solidity, Ethers.js, FAISS, Redis, YOLO, OpenCV, TensorFlow, PyTorch, Docker.
Open to research internships, collaborations, and engineering roles where resilience, scale, and impact matter. 📩