Abhishek Tiwari

Engineered an AI Talent Partner platform using FastAPI, Model Context Protocol (MCP), and Agentic AI Architecture, leveraging Hexagonal (Port-Adapter) Design to enable plug-and-play integration of enterprise tools, HR systems, and knowledge sources, significantly reducing development effort for new capabilities and integrations.
Built a multi-stage LLM-driven Talent Intelligence Pipeline that synthesized employee, peer, and manager feedback, OKRs, IDPs, competency benchmarks, and historical review data to identify skill gaps and generate personalized recommendations across learning, performance, career development, and succession planning, contributing to a 26% reduction in organizational skill gaps.
Architected and built a Java Spring Boot external API platform exposing secure, rate-limited REST APIs for HR services, implementing OAuth2 Authentication, Multi-Tenant RBAC, and Traffic Throttling to enable scalable third-party integrations while isolating core microservices from external traffic spikes and abuse.

Owned high-throughput Spring Boot microservices (100K+ req/day), reducing p95 latency by 35% via Redis cache-aside, connection pooling, and query optimizations with write-through consistency controls.
Built event-driven ILT attendance system using Redis Streams handling 40K+ daily events, absorbing session-start traffic spikes via consumer groups and idempotent processing for reliable tracking.
Designed multi-tenant Global Search using Spring Boot + Elasticsearch, sustaining 20K+ peak QPS with sub-100ms p95 latency via tenant-scoped indexing, custom analyzers, and optimized query filtering.

Built a background processing microservice using Asynq (Redis-based distributed task queue) to offload async workloads from core APIs, improving system modularity and scalability.
Implemented reliable task execution mechanisms including retries, dead-letter queues (DLQ), idempotent processing, and robust failure handling to ensure resilience under high load and partial failures.