Nivedita Shainaj Nair

About

Data/ML platform engineer with healthcare (Cigna) experience owning an end-to-end pipeline spanning Airflow + Debezium CDC ingestion, PySpark/SQL transformations, rigorous data quality gates, and feature-store/API serving for ML training and inference. Worked at 10+ TB scale and cites a ~30% latency reduction plus stronger reliability via idempotent design, monitoring, and backfill-safe reprocessing; also built pragmatic early-stage data pipelines at Frankenbuild Ventures.

Key Strengths

• Owned end-to-end production ML data pipeline (ingestion to serving) in healthcare domain
• Built robust data quality layer (schema validation, null/duplicate checks, quarantine/alerting) to prevent bad data reaching ML models
• Improved pipeline reliability with monitoring/alerting, retries, and task-level reruns for faster recovery
• Handled CDC late-arriving/duplicate data via idempotent design and deduplication using keys/timestamps
• Scaled and optimized Spark/SQL transformations for large volumes (10+ TB platform) and improved runtime/data availability
• Reduced data latency by ~30% and improved feature freshness for downstream ML use cases
• Implemented feature/dataset versioning and training-serving consistency via feature store for reproducible ML
• Delivered early-stage pipeline in ambiguous environment by scoping key metrics and choosing pragmatic batch-first approach

Experience

Education

Skills

Languages