Ankush Banthia

About

Financial data engineer who has built and operated high-volume batch + streaming pipelines (200–300 GB/day; 5–10k events/sec) using AWS, Spark/Delta, Airflow, Kafka, and Snowflake, with strong emphasis on data quality and reliability. Demonstrated measurable impact via 99.9% SLA adherence, major reductions in bad records/nulls, MTTR improvements, and significant latency/runtime/query performance gains; also built a distributed web-scraping system processing 5–10M records/day with anti-bot and schema-drift defenses.

Key Strengths

• Owned end-to-end financial data pipelines (batch + streaming) at 200–300 GB/day and 5–10k events/sec
• Improved SLA adherence to 99.9% and reduced failure rate to <0.5% (from ~3–4%) via retries/error handling
• Reduced MTTR from ~2 hours to ~20 minutes through operational improvements
• Reduced bad records ~40%; cut null rate in critical columns from ~8% to <1%; duplicates near 0%
• Reduced end-to-end data latency from ~20–30 seconds to <5 seconds; batch freshness from T+2 hours to T+30 minutes
• Implemented robust data validation/quarantine patterns (schema enforcement, constraints, anomaly detection, DLQ)
• Performance optimization: pipeline runtime reduced from ~2 hours to ~35 minutes; query performance improved ~5x
• Designed resilient large-scale external data collection system (5–10M records/day) handling anti-bot, schema drift, and backfills

Experience

Education

Skills

Languages

Certifications