Overview
What happens when AI agents can rewrite millions of lines of code orders of magnitude faster than humans can review them? That question drives the Future of Scalable Software Engineering FoSSE team in Microsoft's CoreAI division. We build AI agents that transform some of the world's largest codebases, including Windows and Azure. Generating code is becoming the easy part; knowing whether to trust it is the real challenge.
We combine AI agents with static analysis, testing, and formal verification to make large-scale software transformation practical and trustworthy, and we design tools that spend scarce human attention where it matters most. The team began in Microsoft Research and moved to CoreAI in 2025, and we continue to collaborate closely with Microsoft Research, so research depth meets production stakes every day.
Here are the main problems we are currently working on:
Migration at scale: We work on migrating hundreds of millions of lines of code, across Windows, Azure, Office, and other products, to work with new versions of dependent packages, new compilers and new platforms. Examples include increasing language-standards conformance of code, remediating CodeQL and other static analysis findings, and retargeting services from x86 to ARM.
Language modernization: We move legacy C/C++ and C# codebases to Rust, with validation gates at every step. Differential testing against the original, held-out tests, and specification extraction with formal verification such as Verus proofs mean each migration is backed by evidence, not optimism.
Specification-driven development: We help brownfield codebases adopt specification-driven development by inferring formal specifications for existing code. The specifications are grounded using execution traces and then used to generate concurrency and fault-tolerance tests, find subtle bugs, and ultimately create a validation guardrail for future AI-driven code changes. We are working with formal tools such as TLA+, F*, Lean and Verus in this work.
You will be onsite at least three days a week and learn and grow through working with talented colleagues.
Responsibilities
- Lead the design and delivery of agentic tools and infrastructure for one or more of our problem areas such as migration at scale, language modernization, or specification-driven development.
- Set technical direction for a problem area end to end, from architecture to production adoption.
- Architect systems that let a small expert team drive transformations across hundreds of millions of lines of code.
- Define the validation strategies and quality bars that make agent-produced changes trustworthy at scale.
- Lead cross-organizational efforts with partner teams, such as Windows, Azure, compiler, and security teams, and with the research community.
- Turn ambiguous, fast-moving problems into roadmaps while staying flexible as models improve.
