San Francisco Bay Area
Tech lead for RoboFlow, the ML Workflow Orchestration Platform for Cruise/GM
I transitioned from Tech Lead to Engineering Manager on the Data I/O team, building data ingestion and data export APIs. Data export in particular is a critical tool that allows customers to export labeled training data from Labelbox. We delivered Export V2, migrating from a label-centric to a data-centric exporter with streaming support. About 6 months later, I was asked to manage the Model team, owning model based auto-labeling, model experimentation and model evaluation tools. My primary contribution on this team was Model Foundry, an auto-labeling tool using foundation models.
Labelbox is a ML Platform providing tooling for training data preparation and labeling, as well as model experimentation and evaluation. As an IC, my primary focus was spearheading improvements to the SDK, building developer friendly APIs providing programmatic access to all Labelbox's products.
I worked on the Big Data Platform team, building Workday's Big Data storage and compute platform on HDFS and Spark for Workday Analytics and Reporting product as well as AI/ML teams. My primary areas of focus were Spark compute observability, HDFS storage API service, and workflow orchestration service. During my tenure, we scaled up the platform to support more than 200 Workday customers, storing petabytes of data, running compute on hundreds of nodes per cluster, across 5 data centers worldwide, as well as on cloud infrastructure. We solved many challenges around scalability, high-availability, reliability, observability, and security.