About
Engineer
I build large-scale backend systems, AI-native tooling, and the occasional research-heavy prototype.
I'm Garvit Batra, a Software Engineer (L60) at Microsoft in Bengaluru, working on Outlook and Substrate Mail Services. I graduated from DTU in 2024 with a B.Tech in Computer Engineering and a 9.51 CGPA, and I still carry the same instinct that got me through college: understand the system deeply, reduce the problem to its real constraints, and ship with ownership.
What I work on
My core work sits in large-scale mail infrastructure. I've worked on Outlook service modernization, including the Model A to Model B2 migration, traffic-splitting systems handling hundreds of millions of requests per hour, OData infrastructure, and decoupling legacy mail flows into more maintainable boundaries. I enjoy systems work because it forces precision: scale punishes hand-wavy thinking.
What pulls me in
Alongside production systems, I spend a lot of energy on AI tooling for engineers. At Microsoft, that has meant building things like Smart Coverage Analyser, custom MCP-based workflows, GRAIL-style impact analysis, and internal CLI tooling that compresses the path from idea to implementation. I like tools that make strong engineers faster without making them sloppier.
Research and technical range
I've also worked across applied AI and research: voice automation with Twilio, Whisper, and GPT; data enrichment pipelines; and video summarization research using quaternion transformers, published at IEEE ICPEICES 2024. That mix of backend systems, product engineering, and ML research is a big part of how I think now-I care less about labels and more about leverage.
Skills
- Languages - C/C++, C#, Python, JavaScript, TypeScript.
- Platforms - .NET, Azure, LangChain, PyTorch, distributed systems, API design.
- Comfort zone - backend architecture, developer tooling, AI workflows, debugging at scale.
Outside the job title
I'm based in Bengaluru, I love building tools that make people more effective, and competitive programming still shapes how I approach engineering. I've solved 1800+ problems, reached Specialist on Codeforces, and taught 70+ students through Placewit. I'm concise by default, data-driven when decisions matter, and generally happiest when I'm building something useful.