Hello, I'm
Balachandra Devarangadi
Software Engineer · Applied Scientist · MS CS @ UMass Amherst
Software engineer who likes building thoughtful, playful things.
scroll to take a walk ↓
⛰ About Me
Hi! I'm Balachandra, a graduate student at UMass Amherst who loves turning ideas into polished, working software. I care about clean architecture, thoughtful details, and building things that actually feel good to use.
When I'm not coding, I'm usually outdoors, sketching something, or tinkering with side projects like this one.
By training, I'm a machine learning researcher and software engineer with 2+ years of experience spanning product engineering and applied AI. My M.S. work at UMass focused on LLMs, retrieval, and efficient ML systems, including research on controllable memory for LLM agents, RAG pipelines, and edge inference, I'm drawn to questions about how models behave and how to make them genuinely useful in the real world.
Before grad school, I spent time at Epsilon building backend systems and data pipelines that needed to just work at scale, and more recently I've been a Graduate Research Extern at Microsoft exploring memory frameworks for LLM agents. What excites me most is the space where solid engineering meets real research, that's usually where the interesting problems live.
⛰ Experience
⛰ Projects
…and here are more
ML / AI
Fullstack
Hardware / Embedded
⛰ Wins
🏆
Hackathon Winner
HackUMass · 2025
Halo makes the web safer and calmer, protecting people with photosensitive epilepsy and helping those with ADHD read with ease.
📄
Best Paper Award
IEEE DCOSS-IoT · 2026
"Collaborative Processing for Multi-Tenant Inference on Memory-Constrained Edge TPUs"
🥇
Hackathon Winner
HackHer · 2026
AI-powered vision assistant that helps visually impaired users shop independently by finding items, reading labels, and understanding products through voice.
⛰ Education
⛰ Publications
Task-Aware LoRA Adapter Composition via Similarity Retrieval in Vector Databases
R. Adsul, B. Devarangadi Sunil, I. Nalawade, S. Govindan
arXiv, 2026
Collaborative Processing for Multi-Tenant Inference on Memory-Constrained Edge TPUs
N. Ng, W. A. Hanafy, P. Kadambi, B. Devarangadi Sunil, A. Gupta, D. Irwin, Y. Simmhan, P. Shenoy
IEEE DCOSS-IoT, 2026
Memory Poisoning Attack and Defense on Memory Based LLM-Agents
B. Devarangadi Sunil, I. Sinha, P. Maheshwari, S. Todmal, S. Mallik, S. Mishra
arXiv, 2026
Density Estimation and Crowd Counting
B. Devarangadi Sunil, R. Venkatesh, S. Todmal
arXiv, 2025
Smart Safety Watch for Elderly People and Pregnant Women
B. Devarangadi Sunil, M. Mysore Sampath, S. Pavan BM, S. Shashank , P. Devaki, A. M.
arXiv, 2023
⛰ Skills
hover a skill to see where it earned its keep
Languages
ML & LLM
Distributed Systems
Databases, Cloud & Infra
Frontend
Backend & Data
⛰ Contact Me
I'd love to hear from you — about work, ideas, or a good trail recommendation.
bdevarangadi@umass.edu© 2026 Balachandra Devarangadi · Built with Next.js