Leading multiple efforts within the newly formed Central Efficiency org (Meta's Doge) to achieve efficiency (efficiency as a supply) across the whole ads serving and ML training stacks, so that we can use the saved resources to fund new revenue-generating projects and Meta's other strategic bets.

* Delivered $40+M savings in resources such as compute, storage, etc. with $100+M additional savings on the roadmap.

* Leveraged LLM to create an agent for surfacing efficiency opportunities and efficiency implementation automation.

* Onboard expensive ML training data generation and processing pipelines to Velox to achieve up to 10x compute cost reduction.

* Query optimization for Presto queries at planet scale in a distributed environment.

*Led/coordinated a cross-org/cross-functional team of ~20 members and delivered a pre-production latency/performance evaluation solution for 2,400+ engineers in Google Assistant. It

1) Cuts latency feedback cycle from ~10 days to <1 hr

2) Can detect latency changes as small as 5 ms

3) Processes ~7m Assistant queries per day

4) Prevented ~200 latency issues yearly and reduced production latency outages to 0

5) Won Assistant Tech Impact Award (only 6 awarded out of 76 projects from 2,400+ engineers)

*Led a team of 5 engineers and developed an automated debugging/root-causing solution that

1) Auto-closed ~4,000 non-actionable test failure bugs and provided root-causing information for ~2,000 bugs per year

2) Increased bug fix rate by ~46% and saved ~20 engineer-years per year

Nanometrics was merged with Rudolph and renamed as Onto Innovation in 2019

Image processing