Hari Billa

About

Senior data scientist with hands-on ownership of production ML and GenAI systems across enterprise churn, clinical Q&A, and real-time fraud detection. Stands out for combining strong MLOps discipline with measurable business impact, including $2M+ retained revenue, 10K TPS low-latency fraud infrastructure, and a clinician-reviewed RAG system that improved retrieval accuracy by ~38%.

Key Strengths

• Owned end-to-end production ML systems from experimentation through deployment and monitoring
• Delivered churn model that retained $2M+ in client revenue
• Reduced ML deployment cycles from 2 weeks to 4 days using MLflow model registry
• Cut deployment errors by 25% via standardized A/B testing and MLflow framework
• Built scalable data pipelines processing 50GB+ of daily transactional data
• Designed production fraud detection pipeline handling 10K+ transactions per second at sub-50ms latency
• Strong experimentation and online evaluation discipline with A/B testing tied to business KPIs
• Applied practical model monitoring and retraining for drift management in production
• Shipped clinical RAG application with strong safety controls and clinician-in-the-loop evaluation
• Improved clinical information retrieval accuracy by about 38%
• Cross-functional collaboration with clinicians and business stakeholders, including C-suite dashboarding
• Balanced latency, cost, precision, and safety tradeoffs in high-stakes ML systems

Experience

Education

Skills

Languages