Helping SWEs build cutting edge, proprietary AI via robust and predictable API endpoints• Built 0-1 implementation of React + TypeScript web client and Flask/MongoDB API to support our ML services platform, including user facing endpoints• With Celery & Redis, built task and worker queue to support continuous service evaluation & testing feature (CI/CD but for AI services)• Built LLM benchmarking playground and infrastructure to allow SWEs to evaluate models (for single model, model migration, and multi-model comparisons) on their own datasets, with simplicity and ease• Built continuous document sync pipeline using Dagster, unstructured.io, and Qdrantvector DB that enabled users to easily setup never-stale RAG services• Iterated on frontend UI and backend architecture to support ideation and experimentation in search of PMF, whilst owning engineering decisions

NeevaAI:• Built the initial Golang implementation of Neeva's web summarization result, combining LLMs with search to deliver concise cited answers to queries; helped build/maintain the React result component• Helped build caching & server-sent-events (streaming) middleware to decouple the slower LLM response from the main search response, maintaining latency targets • Built React UI components and Golang code path to support expansion of NeevaAI search result to include query suggestion features to increase engagement• Wrote, eval'd, and iterated on multi-doc summarization prompts for Anthropic's Claude and OpenAI's GPT-3 and GPT-3.5; helped onboard others to prompt engineering processTravel Full Page Experience:• Using React + TypeScript, led frontend implementation and integration of all new components, according to Figma spec; updated parts of component library • Helped with design and implementation of data ingestion pipeline, built with Golang and PySpark, that took in a raw JSON feed and built into structured documents• With design team, often participated in prototype explorations and brainstorming sessions for new features

Diagnosed and solved issues with the technology used in LMU classrooms ranging from personal computers to event space audio/visual systems. Documented issues with ServiceNow ticketing system. Mentored new staff on office procedures, common issues and strategies to find solutions.

Implemented a mix of new features and bug-fixes on 3E, a large financial management product used by ~90% of global top 200 legal firms. Used Angular, TypeScript, NgRx, and RxJS to build front-end features; C# and ASP.NET Core to build backend features. Tested using Jasmine and RxJS Marbles. Worked in remote and agile environment.