JEEVA M
I build AI-nativesystems forresearch andoperations.
Currently designing ResearchOS — an operating system for the complete research lifecycle. Previously at IISc, HCLTech, and Zevaras.
SYSTEMS
What I am building
ResearchOS
In DevelopmentA unified operating system for AI and ML research — combining experiment tracking, paper authoring, knowledge graphs, AI assistance, and offline-first SDK into one connected platform.
ForgeMCU Studio
PrototypeMulti-agent AI platform that generates production-quality embedded firmware — architecture, drivers, testing, static analysis, and build validation — using LangGraph and MCP.
EVIDENCE
Where I have worked
AI/ML Research Intern
2025 — 2025Indian Institute of Science (IISc)
- Developed YOLOv8 Nano segmentation models for precision agriculture — 3.1M parameters, 78ms latency
- Created and annotated dataset of 358 RGB images with 4,812 polygon masks using Roboflow
Software Engineering Intern
2024 — 2024HCLTech
- Contributed to enterprise-scale software development projects
- Collaborated within cross-functional engineering teams
Software Engineer
2024 — 2025Zevaras
- Built production RAG platform for legal document intelligence with hybrid retrieval
- Implemented query rewriting, dense retrieval, BM25, RRF fusion, and cross-encoder reranking
First Prize
2024HCLTech Embedded Systems Hackathon
Built AI-assisted code generation for embedded systems using ESP32 and Arduino. The project demonstrated practical AI application in resource-constrained environments.
METHODOLOGY
Neuromorphic intrusion
detection
Applying Spiking Neural Networks to automotive CAN network security. Detecting intrusions through voltage fingerprinting on resource-constrained embedded systems.
Papers under review
Research conducted at IISc
PRINCIPLES
I design systems,
not just features.
Every project starts with architecture — understanding the problem, designing the system, then building with intention. I believe the best engineering happens before code is written, and the best engineers think in systems that remain understandable as they grow.