Hi! my name is
Building intelligent systems at the intersection of research and production.
I build things that areactually useful.
I'm a Computer Engineering student at Thapar Institute (batch of 2027). Most of my work is applied AI: agent systems, RAG pipelines, and the engineering that keeps them reliable once real users show up.
Lately that has meant fine-tuning a quantized LLaMA with QLoRA and RLHF for the safety layer of a mental-health chatbot, winning Samsung's PRISM hackathon with a Chrome extension that runs all its ML inside the browser, and giving AI coding agents persistent project memory over MCP. Earlier I was CTO at a fitness startup and built the exam datesheet generator my institute still runs on.
Outside of building, I read papers on NLP, multi-agent systems, and LLM efficiency, and keep a steady LeetCode habit (200+ problems so far).
The stack behindthe systems.
44+ technologies across 6 domains. Hover any row to pause.
Hover any row to pause · Each row scrolls continuously
Things I've builtand shipped.
▸ Scroll — the belt advances · Click a card for the full case study
Where I've workedand shipped.
- Led a 5-member team building a fitness platform with three modules: the web platform, LangChain-based nutrition and diet recommendations, and AI workout form tracking. Contributed hands-on across all three.
- Users had no way to check gym form at home, so we used Google's MediaPipe to track body landmarks from the camera and check exercise form in real time.
- Made the institute's exam datesheet generator: a multithreaded scheduler (Flask + MySQL) that cut end-to-end generation time 3x under high constraint density.
- Wrote its constraint solver, which handles 400+ variables (subjects, rooms, faculty) with real-time conflict detection and automated resolution.
- Built a multi-agent safety pipeline for mental health chatbots: emotion, sentiment, and safety modules analyze each input independently, then a deterministic aggregator decides the final response. Tested it against 50+ adversarial prompts for edge cases.
- Fine-tuned a quantized LLaMA with QLoRA and RLHF for safety alignment, layered behind toxicity filters and fallback rules so no single model failure reaches the user.
Recognitionthat followed.
Pediatric Epilepsy Research Symposium
Won 1st place in the Clinical/Engineering category for the oral and poster presentation at 'Advances in Pediatric Epilepsy: From Bench to Bedside', organized by the Pediatric Neurology Unit at PGIMER Chandigarh and the Academy of Pediatric Neurology (AOPN). Presented as part of the Proton BioSystems Lab team (TIET) under Dr. Harpreet Singh.
Samsung PRISM Web Agent Hackathon
Won Samsung's PRISM Web Agent hackathon with KONTA, a Chrome extension that organizes browsing into a searchable knowledge graph using on-device semantic embeddings and Louvain community detection (NDCG@5: 0.91).
Amazon ML Challenge 2025
Placed in the top 100 teams nationwide in Amazon's flagship machine learning competition.
Robowars, Saturnalia
Won the Robowars competition at Saturnalia, Thapar Institute of Engineering and Technology.
Commit history
in the last year.
Let's build
something.
Open to internships, research collaborations, and interesting side projects. Drop a message and I'll reply within 24 hours.
sanyamwadhwa.in@gmail.com