Senior Software Development Engineer
Tech Lead for Agentic, ML Ops, and Data Platform
Amazon Advertising Partner Tech Org
- Architected and delivered Partner Knowledge Base; wrote doc, spun up project, aligned across 6 different tech and product teams, got executive sponsorship, and created a durable pattern for knowledge sharing with Agents across the Partner Tech org. Built on MCP, Strands, AWS AgentCore, and Bedrock.
- Rearchitected org's MLOps platform, cutting time to production from 3 months to 2 weeks and delivered 12 models in 6 months vs the 1 model in 2 years prior. Built serverless AWS Batch infrastructure scaling across thousands of machines, turning annual causal models requiring months of scientist involvement into weekly jobs with less than 1 failure per week across entire infrastructure. Made models CI/CD raising the quality bar and preventing regressions from sneaking in. Trained up scientists on new workflow, both speeding up and making their work more reliable.
- Revived ML Ops initiative solo after it had sat idle for a year. Demonstrated impact within a month, and used its success to establish a permanent AI/ML/Data platform team that has grown to seven engineers. Serve as tech lead and previously acted as interim manager, partnering directly with product and senior leadership on strategy, staffing, and delivery.
- Guided delivery of recommendation, causal, experimentation, and predictive models across Product, Science, and Engineering, translating product use cases into data, model, and serving requirements. Unblocked integration issues, improved model interfaces, and brought scientists who were previously left out into requirements reviews and technical syncs to prevent late-stage pivots.
- Enabled A/B measurement of website product changes for the first time in our org building around an uninstrumented stack and a customer base too small for existing treatment-assignment tooling. Drove end-to-end traceability improvements and partnered with scientists to develop a custom experimentation pipeline.
- Designed and built a generic Spark data precompute pipeline which processes over 1 PB of data each week and has influenced more than $3B in advertiser spend in 5 years. Handles arbitrary queries, transforms, indexes, archival, and federated ownership of datasets.
- Mentored two interns who both earned full time Amazon offers, guiding them through structured projects and teaching durable problem solving methods, not just code. On my team I find and create opportunities for junior engineers to demonstrate growth and senior engineers to take on broader technical challenges.
- Drive a lightweight process mentality on my teams by piloting ideas, measuring their values in retrospectives, and tweaking or removing meetings and reviews which do not improve outcomes. Led operational training and established runbooks and support patterns across two teams.
