Engineering low-latency Java backends, distributed systems and production-grade AI/RAG for multi-tenant enterprise platforms.

Backend engineer at Teamium.io, building a multi-tenant scheduling platform and the AI layer on top of it, from pgvector-powered resource matching to an LLM assistant for schedule queries.

• Building: recommendations, semantic search and RAG over multi-tenant data
• Strengths: Spring Boot microservices, PostgreSQL performance, async and event-driven design
• Foundations: system design, DSA, design patterns, maintainable production code

Languages
Backend & Frameworks
Architecture
Databases & Messaging
AI & LLM
Cloud & DevOps
Testing

_{JAN 2026 – PRESENT · GURUGRAM, INDIA}

Multi-tenant, configurable enterprise scheduling platform managing 20,000+ personnel, equipment and partner resources.

• Talent-AI recommendation engine: Designed and shipped skill-based resource matching on Gemini embeddings + PostgreSQL pgvector (HNSW), returning top matches in 5–10 ms and cutting staffing search time by ~50%.
• Semantic resource discovery: Hybrid search combining vector similarity, keyword matching and tenant-level filters, so planners find crew by skills and availability instead of exact keywords.
• Enterprise LLM assistant: Architected a scheduling chatbot on Spring AI + LangChain4j with RAG pipelines and function calling over multi-tenant data.
• High-throughput ingestion: Excel pipeline that parses and validates 10K+ records per upload, cutting client onboarding from 10 days to under 4 hours.
• Event-driven notifications: Tenant-configurable Email, In-app, SMS and Microsoft Teams delivery on async events + AWS SNS, plus automated recurring scheduling workflows.
• Reliability: Global error-handling architecture; legacy monolithic logic decomposed into reusable modules for better fault tolerance and response times.

Java Spring Boot PostgreSQL pgvector Spring AI LangChain4j AWS SNS

_{JUN 2025 – NOV 2025 · BENGALURU, INDIA}

High-availability enterprise spend-management platform.

• Microservices: Architected core Spring Boot services for expense processing, multi-level approval workflows and travel management.
• Resilient integrations: MakeMyTrip travel APIs with retries, timeouts and idempotent booking handling, preventing duplicate bookings on network failures.
• Non-blocking APIs: Spring @Async, CompletableFuture and AWS SQS for travel calls and notifications, cutting response time by 50%.
• Query performance: Flyway schema migrations plus MySQL execution-plan tuning with targeted indexes and rewrites, taking API response time from 100 ms to 5 ms.

Java Spring Boot MySQL Flyway AWS SQS React.js

_{OCT 2024 – DEC 2024}

Fault-tolerant background-job processing across Dockerized worker nodes.

• Async job processing on RabbitMQ, decoupling producers from long-running background work
• Load balancing: round-robin and least-loaded strategies across worker nodes
• Fault tolerance: retries with dead-letter queues for failed tasks
• Stable throughput under heavy traffic, with workers that scale out as containers