Principal Software Engineer at Netsmart Technologies (2020-03 – Present)
- Architected a human-in-the-loop data validation platform processing 3M+ records/month; engineered expert-matching logic that routes tasks by domain, language, and quality history — directly analogous to Sovrano AI's evaluator assignment system.
- Led design and delivery of RESTful and GraphQL APIs (Python/FastAPI) serving 20+ internal teams; maintained 99.98% uptime and sub-80ms p95 latency at peak load.
- Integrated OpenAI, Anthropic, and HuggingFace LLMs into clinical documentation workflows; built standardized prompt templates, output scoring rubrics, and inter-rater reliability dashboards.
- Managed a team of 5 engineers; reduced production bug rate by 31% through TDD culture, automated regression suites, and structured code review processes.
- Drove migration from monolithic Django app to containerized microservices (Docker/Kubernetes), cutting deployment cycle from 50 min to under 7 min.
Senior Software Engineer at HomeAway (Vrbo / Expedia Group) (2016-08 – 2020-02)
- Owned microservices for real-time pricing and availability across 2M+ listings; built on Kafka event streaming, PostgreSQL, and Redis caching layers.
- Developed an internal annotation and quality-assurance tool used by data science teams to label and review training datasets for ML ranking models — hands-on experience building evaluation tooling at scale.
- Collaborated with ML engineers to instrument evaluation metrics, A/B frameworks, and automated scoring pipelines for search ranking features; improved NDCG@10 by 9% over two quarters.
- Designed OAuth2/JWT authentication, role-based access control, and audit logging for a multi-tenant platform serving enterprise clients.
Research Engineer & Postdoctoral Fellow at UT Austin — Dept. of Computer Science (2013-08 – 2016-07)
Postdoctoral researcher in the Human-Centered AI Lab; built experimental platforms for crowdsourced and expert human evaluation of NLP model outputs.
- Engineered a Django/React annotation platform used in 4 published studies, supporting 1,200+ annotators across 6 languages with configurable task templates and real-time quality monitoring.
- Co-authored 5 peer-reviewed papers on human evaluation methodology, annotator agreement metrics, and feedback-driven model improvement — foundational work in the RLHF research stream.
- Supervised 8 M.S. students; taught graduate seminar on Human-in-the-Loop Machine Learning (CS 395T).
