AI researcher

Raghavendra
Kotikalapudi

I work on multi-step reasoning and reinforcement learning at Microsoft AI, on the MSI team.

Raghavendra Kotikalapudi
Ragha, for short.

Previously

At Google DeepMind, I worked on Gemini’s post-training, from 1.5 Pro through 2.5 Pro and thinking models. As one of the ICPC team leads, I contributed to Gemini’s gold-medal performance at the World Finals.

I also led research on instruction tuning and safety for Bard and Gemini 1.5, and was part of the small team that launched the first version of Bard in 100 days.

Selected
publications

All on Scholar

Highlights

Research in the world.

  • Gemini as a STOC reviewer

    Contributed to the Gemini Deep Think team providing automated feedback for STOC 2026. Over 80% of submitted papers opted in; 97% found the feedback helpful.

  • Gold at the ICPC World Finals

    Gemini 2.5 Deep Think placed 2nd overall in Baku, solving 10 of 12 problems—including one no human team could solve.

Earlier highlights

  • Facebook Civic Hackathon winner

    Won with “Find ’n Park”, a computer vision model that detected available parking spots in real time using Seattle’s open data.

  • How Depressives Surf the Web

    Co-authored a New York Times Sunday Review op-ed about our research linking internet usage patterns to depression in college students.

Patents

Methods for more
efficient language models.