Ankit Kumar

Backend & Platform Engineer with 4+ years building distributed systems, data pipelines, and Kubernetes infrastructure at Flipkart.

Reduced

BigQuery analytics cost by ~70–80% via fk-Juggernaut (ClickHouse on K8s reading Parquet from PVC) and delivered a Central Billing Platform that saves 10–20 engineering hours/week across 11 service providers.

Software Engineer - EnterpriseMinds (Client: Flipkart) - Bengaluru

(2023-12)

• fk-Juggernaut — ClickHouse Analytics Platform (GCS Exports → Kubernetes Storage) (Java, ClickHouse, GCS, Kubernetes, Docker, Helm, Metabase)
• Reduced per-query analytics costs by ~70–80% by migrating Billing dashboards and services off BigQuery scans to a ClickHouse-on-Kubernetes serving layer that reads Parquet from PVC mirrored from GCS exports, eliminating bytes-scanned charges for day-to-day queries.
• Eliminated manual sync ops (100% removal) by building a sidecar with parallel downloads, resume support, change detection, orphan cleanup, and cron scheduling.
• Enabled zero-touch onboarding across 5+ teams by auto-detecting 10+ file layouts and 9 date-partition formats and auto-managing logical view tables (create/validate/drop).
• Hardened deployment with Docker/Helm on fkcloud and migrated auth to Machine Identity via AuthN/AuthZ v3 workload identity pools.
• Central Billing Platform — Invoice & Cost Attribution (CPG Billing/OTM) (Java, Spring Boot, BigQuery, MySQL, Kafka, Prometheus, Metabase)
• Delivered dashboards, automated monthly invoices, and cost alerts across 11 service providers, saving 10–20 engineering hours/week and accelerating cost visibility for engineering leadership.
• Improved billing accuracy by applying custom rate-card logic, anomaly detection, and automated cost rectification across BigQuery billing datasets.
• Automated invoice ops end-to-end via scheduled validation/generation, RBAC-gated distribution, recipient-mapping APIs, and alerts for spikes/dips, attribution gaps, and budget exhaustion.
• Standardized Metabase + ClickHouse analytics; revamped P50/P98/P99 latency dashboards and deployed Prometheus alerts for budget and project-level anomalies.
• In progress: Cutting Dataform pipeline ops from ~1 day → ~10 minutes via runbook automation and audit-table driven alerts; adding Grafana + Zenduty P0 monitoring.
• OTMD / OmniChat — Testing Infrastructure & Integrations (Docker, Helm, Kubernetes, Flow CI/CD, CodeHub APIs)
• Achieved 100% coverage of critical OTMD flows by building hermetic test infra, integration tests, Dockerfile, Helm charts, and CI/CD pipelines; raised unit-test coverage from 0% → 90%.
• Improved security posture by integrating Grayskull runtime secrets, removing plaintext secrets from configs and CI/CD across Billing and OTMD.
• Increased reliability as primary on-call by instrumenting Prometheus with Sev-0 alerts and optimizing Java pipelines to cut latency by ~50% while clearing all Sonar critical issues.

Software Engineer - Stiflink Management OPC Pvt. Ltd. - Remote

(2021-11 - 2023-11)

• Launched a microservices SaaS on AWS Elastic Beanstalk with RDS, rate limiting, and load balancing using Spring Boot, Spring Cloud Gateway, Eureka, and Resilience4J.
• Implemented RBAC (Spring Security/JWT), OAuth2/Okta-secured REST APIs, Twilio OTP, Swagger docs, and JUnit/JMeter test suites — increasing security and delivery velocity.
• Eliminated end-of-life risk by upgrading from Java 8 / Spring Boot 2.x → Java 17 LTS / Spring Boot 3.x with zero downtime.

B.Tech - Computer Science & Engineering - Motihari College of Engineering, Bihar (2017-08 - 2021-11)