I am a senior researcher at Microsoft Research, Redmond, in the AI for Software Systems group led by Dongmei Zhang. My research combines AI, formal methods, and optimization to develop adaptive AI tools and analyze the reliability and performance of cloud systems.

My recent work spans three areas:

• Adaptive AI and Coding Agents: Co-developing model routing for GitHub Agentic Workflows, with a focus on matching models to tasks and learning from past runs to improve routing decisions over time while balancing quality, cost, and latency.

• Cloud Reliability and AI-Assisted Verification: Using LLMs to extract structure and intent from logs and natural-language specifications, then applying formal and data-driven methods to build executable alerting rules, generate high-coverage protocol tests, and learn configuration invariants that proactively detect misconfigurations.

• Heuristic Analysis and Optimization: Reasoning directly over algorithm source code to uncover performance gaps in complex or randomized heuristics, combining code-aware input generation with guided search instead of requiring hand-built mathematical models.

I graduated with a Ph.D. in CS from UCLA in 2022, advised by two amazing people, Prof. Todd Millstein and Prof. George Varghese.

My research has received the ACM SIGCOMM Best Student Paper Award (2020) and the IRTF Applied Networking Research Prize (2023). My dissertation was runner-up for the ACM SIGCOMM Doctoral Dissertation Award (2023).

Before UCLA happened, I was an undergraduate in the CSE department at IIT Kharagpur where I was advised by Prof. Sandip Chakraborty for my undergraduate thesis.

Click here for my Microsoft webpage.

Contact Me

Office:   Microsoft Building 99, Redmond, WA
Email:   sivakakarla microsoft com