Will Chen

• Pipeline Platform: Built the reusable pipeline platform for a USDA federal contract, shipping 50+ end-to-end ETL pipelines from S3 to Amazon Redshift in Python and AWS Glue; owned 150+ pipelines across the team and cut new feed onboarding time by 50% through reusable templates other engineers built on top of
• Runtime Wins: Killed recurring Glue job timeout failures by re-architecting multi-step workflows with AWS Lambda and Step Functions, partitioned loads, and tuned DPU allocation, cutting end-to-end pipeline runtime by 60%
• Delivery Infrastructure: Built the deployment system engineers used daily with GitHub Actions CI/CD, Docker container builds, and Terraform-managed AWS resources across 200+ tables in 3 environments, taking deploys from ~1 hour of manual work to ~10 minutes and eliminating unplanned schema rollbacks
• Catching Bad Data Early: Wrote data quality checks into AWS Glue jobs that fail-fast on whitespace, type mismatches, and malformed records; automated schema drift detection caught 3 upstream schema changes over 6 months before any bad data reached USDA analyst dashboards
• Platform Observability: Instrumented 150+ pipelines in Datadog with SLA tracking, volume anomaly detection, and on-call alerting, giving the team a single pane of glass into pipeline health instead of grepping through Glue logs
• Unblocking Analysts: Shipped self-serve Tableau dashboards and written runbooks serving 100+ users at USDA, powering congressional reports and internal budget reviews, and saving the team ~2 hours/week of ad-hoc report pulls

• Protein Data Analysis: Lifted unknown protein classification accuracy from ~70% to 89% by training scikit-learn ensemble models on mass spectrometry data, serving 10-30 researchers, scientists, and doctors across 10+ collaborating labs in bacterial sample analysis
• Automation & Reliability: Cut 2+ hours of daily manual work by scripting instrument data processing in Python; kept the 5TB research database intact with automated weekly backups and upload validation that catches corrupted writes before downstream analysis

Open Source Contribution: Contribute to Isaac Lab, NVIDIA's open-source GPU-accelerated framework for reinforcement learning and imitation learning across multiple robot embodiments, building modular Python simulation environments for manipulator and AMR tasks in PhysX