rajatmehta03

Built and deployed scalable ultra low latency order streaming system handling ∼5K orders/sec from Matching Engine using Aeron Archive for microsecond-latency publishing, and with dual-consumer architecture routing data to Kinesis → S3 → Snowflake for persistent storage (6.48TB/month, 12-13B rows) and Redis Streams for real-time WebSocket delivery to users. Architected event-driven settlement system reducing settlement time from 5 minutes to real-time, eliminating user-facing margin/collateral balance discrepancies through streaming architecture (Wallet Core Stream → BullMQ aggregator → settlement queue) with full audit trail and transaction-level traceability. Resolved critical database performance bottlenecks across 320M+ row tables through range-based partitioning, aggressive auto-vacuum tuning(to prevent CPU spikes), and hot-cold table separation, reducing query latency from timeout to double digit ms. Built multi-collateral trading system supporting deposits in any token with USDC margin allocation using event driven architecture with BullMQ, implementing idempotent operations for deposits, transfers, settlements, and liquidations. Developed multiple platform features: Points Distribution (trades/liquidity rewards), Referral Payouts, Grid Trading Strategies, Daily platform analytics.

Design and built numerous user onboarding flows backend to enhance user engagement and retention. Implemented personalized email communication with custom autogenerated screenshots for clients. Established end-to-end Keychain infrastructure, including code setup, dockerization, github actions, AWS deployment, and VPN. Deployed an organization-wide CDN solution for images, significantly reducing latency and enhancing user experience. Successfully managed the migration of millions of product images to the CDN. Developed event capturing pipelines using Rudderstack.

Designed and built complex billing service from scratch based on publish-subscribe model that do the billing for all exchanges in 2.5-10min for 3.5L+ clients. Tech Used: Golang, Pyspark, Kafka, AWS(ECS, ECR, Step Fn, S3, Lambda, Glue, Mysql Aurora DB, SNS,Event Bridge etc), Docker, PagerDuty, Terraform. Automated the manual files downloading work by building a organisation wide File Download Service in Golang/Python, which download the file from exchange portal by web scrapping and FTP server, and save it to S3, this service is deployed on AWS ECS. Developed a microservice in Golang to show the billing charges upfront before user place order on trading app with upto 10ms response time. Worked on Big Data ETL jobs that handles GB’s of data daily using Pyspark , and deployed on AWS GlueJob. One of the data migration which i have implemented saved 75L Rs of a company. Built a service to generate alerts for suspected traders trading outside India by using trader IP addresses using AWS Step Fn, Lambda, Event Bridge, Glue, Signed URLs. Worked on User Onboarding Journey, Recommendation, and Advisory which helps in onboarding new users and recommends different types of stock lists, which helps them in getting started, across different segments like ETF, Options, Equity, Single stock. Design and built BFF, which helps in pushing all the dependency on the backend for UI-related stuff, resulting in less load on the frontend, fewer app build and release, and allowing product teams to experiment with different configuration and allowing them to customize as per user behaviors. Handles RPS: 17K.

Health Monitor App: Developed a backend system to retrieve user data such as heart rate, o2 level, temperature, and BP from Google Fit APIs. Analyzed the data in batches to notify users of potential health issues. Enabled users to view trends in their heart rates and o2 levels on a daily basis. Implemented Google-based authentication. Indian BDIP: Internal Project which imports all the stocks of employee and notify whether they have bought stocks which are restricted by there employer. Tech Used: Spring Boot, JPA, Spring Security, and used Role based authentication.