Anay P.

About Anay P.

A software engineer at Meta, I build the infrastructure and orchestration layers that transform LLMs at large scale into autonomous Agentic Developers. Currently at Meta, I lead the integration of AI-driven workflows for the AOSP (Android Open Source Project) repository, collaborating with Meta SuperIntelligence Labs (MSL) to bring agentic capabilities to one of the world\'s most complex codebases.My work focuses on optimizing the developer inner loop by bridging the gap between raw model power and production-ready execution. I have led the end-to-end development of AI-native products from inception to production—building custom agentic orchestrators and specialized toolsets that allow LLMs to independently read, write, build, and test code within the IDE.I have significantly improved the quality and reliability of GenAI/LLM-based tools by creating large-scale benchmark and evaluation systems, including Meta SWE-Bench, for agentic platforms. By establishing these high-fidelity measurement frameworks and seamless context-delivery systems, I have defined the standards for autonomous AI-human collaboration in enterprise software engineering.

Experience

1. Software Engineer

   Meta

   Apr 2024 — Present · Bellevue, WA, US

   Dev Infrastructure - Developer Experience Building Meta\'s GenAI-powered developer tools and infrastructure for Reality Labs and AOSP — leading the design and delivery of agentic platforms, AI code authoring tools, and large-scale migration systems that are transforming how thousands of engineers ship software. Agentic AI Platform (Devmate)Architected and built Devmate, Meta\'s LLM-powered agentic developer platform for AOSP. Designed scalable multi-agent workflows with static and dynamic context enrichment, and stood up the full Evals, Graders, and benchmarking infrastructure to measure LLM code generation quality. Core areas: Context Engineering, Prompt Engineering, MCP, RAG, and distributed agentic architecture. Agentic UI EngineerBuilding an agentic system that converts design inputs — Figma mocks and screenshots — into production-ready frontend code diffs, enabling developers to ship UI changes end-to-end with minimal manual intervention. Bug2DiffSolved the Bug2Diff workflow for Oculus Chromium browser bugs using an AI pipeline powered by Confucius and Claude Code, scaled on Meta\'s internal infrastructure — achieving automated bugs fixes for browser team. APEX Migration for AOSPLeading a pod driving the APEX migration for Android OS — a complex, industry-scale migration. Designed and implemented multiple long-running, large-context agents to automate the effort. AI Code Authoring ToolsDeveloped AI-powered code authoring tools leveraging Llama for Visual Studio IDE — including Auto Complete, LLM-based Chat, and codebase interaction — built end-to-end on the Language Server Protocol, VSSDK, and.NET. Developer Infrastructure- Led Linux developer experience improvements at Meta- Optimized build systems via Buck2 remote execution and CI pipelines (Translator to Citadel migration)- Owned source control, CI/CD, landing, and release tooling for Reality Labs (AR/VR)

Education

• Kalinga Institute Of Industrial Technology

  Bachelor of Technology (B.Tech.), Computer Science and Engineering

  2012 — 2016

• Arizona State University

  Master of Science - MS, Computer Science

  2018 — 2020