Nachiketa Raina

About Nachiketa Raina

I build distributed systems that need to be fast, reliable, and scalable.Currently at Amazon as a Software Development Engineer, where I own mission-critical services processing 10M+ requests daily with sub-5ms latency. My work focuses on delivery prediction systems, ML platform engineering, and backend infrastructure that powers global e-commerce.Technical expertise- Backend: Java, Python, Scala, Spring Boot, Node.js- Cloud: AWS (EMR, Lambda, Step Functions, DynamoDB, S3)- Data: Apache Spark, Hadoop, petabyte-scale pipelines- Architecture: Microservices, event-driven systems, ML deploymentRecent highlights:→ Built ML platform improving prediction accuracy 2.4%, driving $360M revenue impact→ → Re-architected Finite State Transducer (FST)–based transit-time models, reducing graph sizes by ~97%(900MB → ~20MB) and restoring P99 latency and memory stability→ Maintained 100% uptime during 3x peak traffic (Black Friday/Cyber Monday)→ Reduced infrastructure costs by $200K+ annuallyI enjoy the challenge of making complex systems simple, fast, and reliable. From algorithm optimization to infrastructure design, I focus on engineering solutions that create real business value.Open to discussing distributed systems, ML infrastructure, performance optimization, and backend architecture.

Experience

1. Software Engineer

   Amazon

   Aug 2022 — Present

   Transit Time Calculation Service - Own service processing 10M+ daily requests with sub-5ms latency for $100B+ GMV - Calculates shipping duration for 200K+ routes (zip-to-zip) for third-party sellers - Built distributed data pipeline ensuring zero customer-facing disruptions2. ML-Based Transit Time Optimization Platform - Improved delivery prediction accuracy by 2.4-2.9%- Drove $360M revenue increase and $29M operational savings - Reduced ML model training and deployment time from 1 week to 10 hours3. Multi-Region Distributed System - Supports 4 calculation models across global markets - Reduced data refresh from 2-3 weeks to 1 day - Compressed storage by 97% while improving route coverage by 30%- Applied Finite State Transducer compression optimization algorithm safeguarding TransitTime Service from having a degraded performance4. Self-Service Configuration Platform - Enables 20+ business teams to manage carrier schedules independently - Reduced operational turnaround from 2 weeks to 1 day - Handles 500+ monthly configuration changes with zero incidents5. Infrastructure Cost Optimization - Implemented S3 lifecycle policies:$100K+ annual savings - Migrated EC2 to serverless: 30% cost reduction - Total savings:$200K+ annually across 10+ microservices6. Operational Excellence & Peak Traffic Leadership - Maintained 100% uptime during Q4 2023 with 3x traffic volume - Led Java 8→17 upgrades and Graviton migration across 10+ services - Expanded logistics holiday service to EU/JP regions

Education

• Maharaja Surajmal Institute Of Technology

  Bachelor of Technology - BTech, Computer Science

  2017 — 2021