HEMANTH KARI

E-Commerce Platform

• ◆Expanded the order fulfillment pipeline using Kotlin, Kafka, and DynamoDB to build event-driven microservices that process over 50,000 orders daily, cutting order dispatch time by 25% during peak holiday periods
• ◆Built storefront sections with Next.js 14, React, and TypeScript, delivering server-side rendered product pages and checkout flows; SEO rankings improved and Largest Contentful Paint (LCP) increased by 40%
• ◆Enhanced inventory synchronization by implementing SQS and Lambda for real-time updates across three fulfillment centers, and introduced Redis-based distributed locking to prevent overselling
• ◆Developed React search components integrated with Elasticsearch, adding debounced autocomplete, faceted filtering, and infinite scroll, which lowered p95 latency from 450 ms to 120 ms
• ◆Created a testing environment using LaunchDarkly feature flags and A/B testing, enabling 15+ checkout experiments that boosted conversion rates by 8%

Digital Auto Insurance

• ◆Led the move to microservices: Came up with a plan for breaking up a big Django project on AWS EKS into smaller, domain-driven services. Moved the team off manual weekly deployments by building a fully automated pipeline with Terraform and GitHub Actions. This shift let us release daily and cut deployment errors by 40%.
• ◆Redesigned the insurance quote wizard using React and TypeScript. The new interface reduced form abandonment by 35% and drove a 28% jump in mobile conversions.
• ◆Solved N+1 query issues and added missing indexes to the main API. With these changes, the p95 latency dropped from 800ms to 180ms, which led to more customers buying from us.
• ◆Built observability from the ground up by making bespoke dashboards and PagerDuty notifications to keep an eye on Prometheus and Grafana. It used to take 30 minutes to find faults, but now it only takes less than 5 minutes.
• ◆They made the CI/CD transition. They used Terraform and GitHub Actions instead of starting it up by hand. Before, it just came out once a week. Now, it comes out every day, and you may roll back automatically.
• ◆Made security better: This is how to set up OAuth 2.0/JWT login with tokens that only work for 15 minutes.
• ◆Coached two junior engineers in weekly pairing sessions, raising test coverage from 65% to 92% and reducing post-release issues by 45%

Cloud-Based Business Platform

• ◆Developed a provisioning engine in TypeScript that autonomously managed the lifecycle of 500+ VMs, reducing manual effort by 90% and cutting setup time from two hours to 10 minutes
• ◆Scaled the platform to support 50,000 daily users by implementing circuit breakers, retry logic, and graceful degradation in Python and Spring Boot microservices, achieving 99.5% uptime
• ◆Optimized database performance by adding connection pooling, read replicas, and targeted indexing in PostgreSQL and MongoDB, lowering p99 latency by 38%
• ◆Led on-call incident response as the designated DRI, creating runbooks and automated remediation scripts that reduced MTTR from 90 to 50 minutes (45% improvement)
• ◆Created OpenAPI documentation for 30+ endpoints and defined an API versioning strategy, enabling partners to integrate 60% faster, reducing onboarding time from two weeks to five days

Research Computing Backend

• ◆Develop a Django/PostgreSQL backend with CI/CD pipelines, supporting over 1,000 concurrent users and maintaining 99.9% uptime; implement caching and query optimization to improve page load speed by 40%
• ◆Redesign the batch processing pipeline to use asynchronous workers and parallel execution, reducing processing time by 81% (from four hours to 45 minutes)
• ◆Create React dashboards that pull data via Prisma ORM for over 2,000 faculty members, simplifying data access and lowering support requests by 35%