◆ Building a role-aware RAG based AI chatbot using AWS Bedrock with Python Lambda orchestration and DynamoDB backed session history to deliver context-aware responses.

◆ Optimizing VectorDB indexing of 1M+ records of S3 using custom Lambda triggers to chunk and embed data using Titan, and store them in AWS OpenSearch reducing service costs from $50K to $20K/year.

◆ Developing data pipelines using 4 Lambdas triggered by EventBridge to store 10K+ security records/day in AWS S3 with IaC provisioned through AWS CDK in TypeScript for downstream LLM workflows.

◆ Built and deployed 12 Python AWS Lambdas using API Gateway, leveraging DynamoDB as global storage for multi-tenant config management, eliminating 30GB data across 20+ Kubernetes clusters and saving $2K/year.

◆ Increased reliability of long-running AI workflows by enhancing the core Java Spring Boot Temporal service, unifying error handling for 59 activities into 4 standardized types and reducing manual retries by 98%.

◆ Built non-blocking customer usage logging in a FastAPI-based AI service for LangChain and Amazon Bedrock, streaming to CloudWatch Logs for real-time billing and cutting latency by 120ms.

◆ Developed a Python AWS Lambda with Cognito post-signup triggers to sync 100% of SSO users (Okta, Microsoft, Google) in real time to on-prem DBs via a custom REST API.

◆ Owned integration testing with real mock events for Lambdas and Spring Boot services, boosting test speed by 40%.

◆ Led the migration of core micro-frontend from React v14 to v18, fixing legacy bugs, preventing regressions in prod.

◆ Optimized application architecture with Redux, and load distribution, cutting bundle size from 300MB to 3MB using Webpacks and reducing load time from 12s to 3.5s.

◆ Boosted UI responsiveness 400% with infinite scroll, lazy loading, delivering components 5× faster using Material-UI.

◆ Developed mock APIs in Node.js for E2E testing, enabling faster prototyping and reducing QA cycles by 30%.

◆ Built 2 Java Spring Boot microservices to ingest, validate, and serve 2M+/yr medical-diagnosis records.

◆ Streamed real-time data from a Spring service via RabbitMQ to the notification service and 2 Kafka topics for the ML team’s analysis and ETL processes, reducing failures and data loss by over 10GB+ during peak stages.

◆ Optimized data retrieval with Redis caching, reducing database calls by 65% and latency by 2s/API request, while aggregating 100K+ user data from PostgreSQL and reports data from MongoDB.

◆ Developed integration tests using Mockito, increasing code coverage by 46% and reducing prod bugs significantly.