Narendra Kumar

Joined as Engineering Manager in 2022; promoted every ~1.5 years, currently Associate Director as scope expanded across Data & ML Platform, org leadership, platform strategy, and productized data/AI systems.

Leadership & Strategy

• Head the Data & ML Platform charter, leading 5 pods and 34 engineers across data infra, platform, analytics, reporting, and ML platform.
• Built a multi-pod platform operating model with clear charter ownership, planning cadence, execution reviews, platform success metrics, and leadership mechanisms across engineering managers, tech leads, and senior ICs
• Defined and executed Razorpay's data platform strategy, enabling AI-driven financial reporting and merchant insights products used by millions of customers monthly.
• Managed cross-functional initiatives across developer productivity, cost optimisation, data availability, data reliability, and platform adoption.
• Designed and implemented organisation-wide mentoring programs benefiting 128 individual contributors across engineering levels

Platform, Product & Business Impact

• Led Data Platform re-architecture, reducing annualised Data Platform Cost by $2M while improving platform availability to 99% and reducing recurring data quality issues by 97%.
• Scaled the platform to 3.6 PB+ storage and 56B+ events/day ingestion with 10-minute data availability for critical analytical workloads.
• Scaled Trino-based query infrastructure to support 50M+ queries/month across analytics, reporting, and self-serve data exploration.
• Led ADA — Anomaly Detection & Alerting, a multi-tenant real-time anomaly detection and fraud-prevention platform built on Kafka, Flink, and ClickHouse, delivering ~80% cost reduction versus Pinot + ThirdEye, sub-30-second detection latency.
• Re-architected Merchant Reporting Platform to generate 523K+ financial reports/month, reducing operational cost by 74% while maintaining 99.99% SLA.
• Built a real-time streaming architecture capable of handling 500x traffic surges during high-volume events such as IPL, ensuring zero downtime for critical data flows.
• Envisioned and launched two platform-led products: ‘DataSync’, a no-code merchant data integration product, and InsightX an AI-driven merchant insights platform.
• Built platform capabilities for feature engineering, real-time data availability, and AI-driven merchant insights, enabling faster experimentation and productionization of ML/data products.

• Built and managed a 36-member Data & DevOps organisation spanning data platform, ingestion, infrastructure, and security.
• Built and led 3 teams that delivered a self-serve data ingestion platform processing 12B+ events.
• Migrated the entire data infrastructure from AWS to GCP, reducing operational costs and improving performance for mission-critical data pipelines.
• Managed infrastructure security setup, including firewalls, WAF, and OWASP-aligned controls, ensuring compliance and security across platform layers.

• Directed a team of 9 engineers to optimise the data ingestion layer, reducing costs by 60% month-over-month.
• Scaled analytics platforms to support 80+ analysts, delivering 50K+ queries daily.
• Optimised Redshift performance, reducing p90 query latency from 22 minutes to under 10 minutes.

• Designed and developed a high-scale data ingestion platform processing 48B+ events daily, with a p95 turnaround time under 10 minutes.
• Built self-serve data platform capabilities for 50+ users, supporting 3K+ pipelines and 10K+ analytical models.
• Deployed and optimised Presto, Hive Metastore, and Metabase for 200+ analysts, enabling efficient querying and reporting.

Joined as Sr. Engineer and progressed through annual promotions to Architect while building enterprise data, AI, and streaming platforms.

• Designed and built Piperr™ - AI for Enterprise DataOps platform, from 0→1.
• Built real-time streaming applications for fraud detection and data enrichment.
• Implemented a cross-category recommendation engine using four algorithms, delivering 500K+ recommendations/day.
• Built and trained a chatbot for flight and hotel booking, including named entity recognition for destinations, dates, names, and domain-specific keywords.

• Estimated, designed, and developed ETL systems for healthcare data processing use cases.
• Received Client Delight Award in 2015.
• Received the Aryabhatta Innovation Award in 2014.

• MovoGrid: Developed with Microsoft Kinect and QPainter, MovoGrid is an interactive, gesture-driven advertising platform for indoor use.