Aman Gupta

Leading development of subscription, entitlements, and metering systems designed to support multi-region, multi-tenant deployments, enabling compliance with regional data laws and reducing outage impact in failover scenarios. Designing revenue-aligned access control workflows that streamline billing accuracy, reducing manual reconciliation time for finance teams by 30%+. Architecting infrastructure changes to support cross-region data replication and high availability, targeting 99.99% uptime for mission-critical services.

Built and scaled backend microservices in Python, Go, and TypeScript powering communication, AI, and healthcare workflows used by 50,000+ medical professionals. Led development of a multilingual LLM-powered medical assistant, implementing LangChain pipelines, PGVector indexing, and semantic search, increasing medical content retrieval accuracy by 70%. Optimized real-time streaming pipelines (FFmpeg, WebSockets, gRPC), cutting end-to-end latency by 50% and reducing streaming failures by 40%. Designed an advanced hybrid search system combining semantic search + recency scoring, boosting query relevance by 80% based on offline evaluation + user feedback. Revamped CI/CD pipelines (Docker, Kubernetes, Bitbucket), reducing build time from 12 min 6 min (50%) and image size by 60%, accelerating deployment cycles. Improved deployment stability via Docker, Kubernetes, structured logging, and automated infrastructure workflowsreducing debugging time by 50%. Identified and fixed a deadlock in an open-source IM server, increasing message delivery reliability and stability for production users. Recognized as Winner of Docquitys AI Productivity Challenge for delivering high-impact AI features.

Implemented RBAC-based user management and automated JWT key rotation, improving security posture across services. Built a reusable pagination framework used by multiple teams, improving API consistency and reducing duplication. Contributed performance optimizations across microservices, improving average response time by 2030%.