Roshan Agrawal

Designed and built a high-throughput audit logging microservice in Java Spring Boot on an event-driven architecture to ingest, order, and persist customer lifecycle events across 100+ Kubernetes clusters spanning multiple datacenters Scaled concurrent RabbitMQ consumer ingestion with a multi-threaded design using BlockingQueue and ConcurrentHashMap, increasing throughput 5x under production load Solved distributed consistency under network latency with background thread scheduling and a 10s SLA, persisted data in PostgreSQL, and used Redis as a distributed cache

Software Development Engineer II at Tata Communications (2024-07 – Present)

• Designed and built a high-throughput audit logging microservice in Java Spring Boot on an event-driven architecture to ingest, order, and persist customer lifecycle events across 100+ Kubernetes clusters spanning multiple datacenters
• Scaled concurrent RabbitMQ consumer ingestion with a multi-threaded design using BlockingQueue and ConcurrentHashMap, increasing throughput 5x under production load
• Solved distributed consistency under network latency with background thread scheduling and a 10s SLA, persisted data in PostgreSQL, and used Redis as a distributed cache
• Built scalable platform REST APIs and backend modules for Vector Databases (Milvus, Qdrant) in a 3-tier microservice architecture across multi-cluster infrastructure, contributing $2M platform revenue
• Worked across the Data Platform stack (Kafka, ClickHouse, Airflow) and benchmarked ClickHouse against AWS Redshift from 1GB to 50TB, beating Redshift by 60% in ingestion and 10x in query time, winning JLR as a customer
• Built end-to-end observability and monitoring with Prometheus, Alertmanager, and Grafana, cutting MTTD from 30 min to 10 min, and owned production support for deployments, networking, and migrations
• Packaged and deployed services with Docker, Kubernetes, Helm, and ArgoCD. Wrote JUnit tests, used GenAI tools for boilerplate under human review, and delivered via GitHub Actions CI/CD and Git-based version control

Intern at Uber (2023-07 – 2024-06)

• Developed agent ticket BI ETL data-pipelines using Python and SQL with cron scheduling, validated via YAML config, and deployed to production, saving 1200 man hours and cutting costs by $12K
• Audited 40+ CRM metrics across BigQuery agent ticket tables, designed ER diagram in LucidChart, created DB schema, validated data coherency, and consolidated a production SSOT table improving team efficiency by 13%
• Queried production datasets using Spark SQL, automated weekly reporting workflows in Python, and ran deepdives and metric analysis that improved CommOps agent performance by 11%

Bachelor of Engineering in Electrical and Electronics Engineering – Birla Institute of Technology and Science Pilani (2024-05)