Lena
Sato
Staff Machine Learning Engineer
I build large-scale ML systems that reach hundreds of millions of people. Over the past decade I've taken models from research notebooks to production infrastructure — training, serving, and the unglamorous reliability work in between. I care about systems simple enough to reason about and fast enough to disappear.
Selected Work
Atlas Serving Stack, Tech lead · 2023
An inference runtime that cut p99 latency 3× for a 70B model. Open-sourced; now powers serving at three other labs.
View →Featherweight Quantization, Author · 2022
A post-training quantization method that holds accuracy at 4-bit. Paper + library, 4k★ on GitHub.
View →Looking Glass, Creator · 2021
An interactive tool for visualizing attention inside transformer layers.
Experience
Staff ML Engineer, Northwind AI
2021–presentLead the training infrastructure for a 70B-parameter foundation model. Cut training cost 40% with a custom data pipeline and brought time-to-first-token down 3× in production serving.
Senior Software Engineer, Meridian
2017–2021Built the recommendation system powering the home feed for 200M+ daily users, lifting engagement 18%. Owned the online feature store end to end.
Machine Learning Engineer, DeepVision
2014–2017Shipped the company's first on-device vision model for real-time object detection on mobile, running at 30fps under 50MB.
Education
M.S., Computer Science
2012–2014Stanford University · Focus: Machine Learning
B.S., Electrical Engineering & CS
2008–2012UC Berkeley
Skills
- /Python
- /PyTorch
- /JAX
- /CUDA
- /Distributed Training
- /Triton
- /C++
- /Go
- /Kubernetes
- /TensorRT
- /MLOps
- /Transformers
Interests
Bouldering, Generative art, Espresso
Awards
- NeurIPS Best PaperNeurIPS, 2019
- Patent: efficient post-training quantization2020
- Internal Engineering Excellence AwardMeridian, 2018