~/riyan $
CS Student at UT Dallas
Full-Stack Developer · ML Enthusiast
01 / about
Building at the intersection of ML, systems, and full-stack engineering. Passionate about shipping things that actually work in prod.
02 / achievements
Scene graph generation using transformers + GNNs for real-time video. 92.16% reduction in training loss across 40,000+ samples.
ML app detecting Parkinson's, Alzheimer's & MS via smartphone camera. Built with PyTorch, MediaPipe, OpenCV + Flutter.
Routed queries to specialized sub-models and aggregated responses via average embeddings. Won first place out of all competing teams.
Led 15 officers. $2,500+ in prizes, 100+ applicants, MLH partnership, judges from Visa and Amazon.
03 / skills
First language I learned — unlocked ML and scripting for me.
YOLO object detection pipeline, Flask APIs, data analysis scripts.
04 / projects
Collaborative coding interview platform with real-time rooms, microservices (central · RTC · worker), and live code execution.
Hackathon AI — breast cancer Q&A with Gemini API, FAISS vector search, and multi-source scraping. Deployed on Vercel.
Assistive tech for visually impaired — fine-tuned Llama 2 + STTran scene graphs + G-Retriever RAG. Trained on 2×H100 GPUs.
Hierarchical AI routing queries to specialized sub-models, aggregating via average embeddings for max accuracy.
05 / contact
Open to internships and research opportunities.