I'm building agents that actually do things - from code reviewers to deep researchers to bash automation. Currently doing my MS in Computer Science at Arizona State University while working as an SDE at MAQ Software (Microsoft Partner).
- π MS Computer Science @ ASU (GPA: 3.93) - May 2025
- π B.Tech IT @ NIT Jalandhar - May 2023
- πΌ Software Development Engineer @ MAQ Software (Microsoft Partner)
- π§βπ« TA for CSE 578: Data Visualization @ ASU
AI/ML & LLMs
Cloud & Infra
Backend & APIs
- AgentLens - 6 agents that code-review other agents. slop detection, security analysis, token optimization. uses MCP servers to fetch live docs. also built automatic prompt optimization into it.
- Multi-Agent-Deep-Researcher - give it a topic, it goes deep. web scraping, PDF processing, entity extraction across dev tools, products, finance.
- Computer-Use-Agent - LLM that runs bash commands (carefully). NVIDIA Nemotron + LangGraph with human-in-the-loop.
- serverless-smart-inbox - email sorting with sentiment analysis. S3 β Lambda β Comprehend β SQS.
- MCP protocol and making agents actually useful beyond demos
- Token optimization (got 4000 β 2500 under 10x load)
- The gap between "cool agent demo" and "production agent that doesn't break"