My Work at Meta
Hardware engineering work at Meta — validating and debugging custom AI silicon, deploying it at scale.
During my four years at Meta, I supported end-to-end validation and bring-up for four custom AI accelerator ASIC generations, from pre-silicon through production deployment. I owned clock, reset, and boot validation across all phases, building system-level infrastructure that scaled from emulation to production and encompassed PCIe, power management, memory, and telemetry subsystems. I caught critical defects like a DMA-to-memory mismatch before silicon was brought up and debugged production escalations involving complex power and clock-management issues. I built, automated, and scaled C++/Python validation frameworks with 500+ test cases, cutting bring-up time from weeks to days.
I defined validation strategy and implemented power-management features that enabled characterization and became reusable frameworks for future ASIC programs. I established stability-testing methodology that gave leadership confidence in silicon stability for fleet deployment. I truly embodied the "AI-native engineer" mentality, leveraging personal AI tooling for code generation and automation analysis, and building productivity tools adopted across multiple teams. I served as the primary validation contact for external partners and mentored new members and transition teams to take over validation ownership.
To learn more about the programs I worked on at Meta, check out the video and links below:
Blog post from Meta on the latest MTIA chip: MTIA Roadmap
The first MTIA project I worked on from start to finish: MTIA Next-Gen