Dhairya Gajjar

About

Backend/ML engineer with healthcare domain experience building secure Medicare/Medicaid data APIs and real-time patient risk scoring. Shipped an end-to-end ML pipeline (scikit-learn/XGBoost) served via SageMaker and integrated into Flask APIs, with strong production reliability practices (Kafka schema validation, regression replay, observability, drift monitoring, and human-in-the-loop guardrails).

Experience

Education

Key Strengths

• Built secure healthcare APIs for Medicare/Medicaid data on AWS (EKS + Kafka + Flask/GraphQL)
• Debugged silent data integrity issues by adding Kafka-consumer JSON schema validation and a message replay regression suite
• End-to-end ownership of real-time patient risk scoring (data processing, feature engineering, model training, deployment, monitoring)
• Latency optimization for real-time ML inference (feature precompute, caching, right-sizing SageMaker endpoints)
• Prevented training/serving skew by reusing identical preprocessing code in training and inference
• Designed reliable multi-step workflow with explicit success/failure states, retries/backoff, fallbacks, and manual-review escalation
• Production Python best practices: layered architecture, Pydantic validation, async/Celery, pytest, observability (logs/metrics/alerts)
• Relational modeling and SQL performance tuning (query plan analysis, composite indexes, avoiding SELECT *)
• AI production guardrails: logging inputs/outputs/latency/confidence, drift monitoring, regression tests, human-in-the-loop for low confidence