Open to opportunities

Nikhil Kumar Agarwal

@nikhilkumaragarwal

Message

I build scalable backend and AI platforms with reliable, low-latency systems.

India

Message

What I'm looking for

I'm looking to build scalable backend, distributed, and production AI systems where I can improve reliability, latency, cost efficiency, and evaluation quality.

At ShyftLabs, I build production backend and AI platforms that process real-time workflows reliably at scale. I scaled NestJS, Redis, and BullMQ pipelines to handle 12K+ jobs daily while sustaining p95 latency under 1.8 seconds.

I’ve implemented multi-tenant RBAC across more than six microservices, productionized Kubernetes and Helm services with OpenTelemetry, Sentry, and OIDC, and reduced operational friction through structured tracing and authorization controls.

I also designed a LangGraph multi-agent platform with 10+ agents, improving internal task completion by 28%. My multi-provider LLM platform across OpenAI, Anthropic, Vertex AI, and AWS Bedrock reduced LLM spend by 32% through routing and caching.

Beyond my work at ShyftLabs, I built KagazCheck, a document intelligence platform supporting OCR validation across 16+ document types, and Grocky, an AI-powered interview preparation platform that improved engagement by 40%.

Experience

Work history, roles, and key accomplishments

Current

Software Engineer

Current

ShyftLabs India

Jun 2026 - Present (2 months)

Built and scaled NestJS + Redis + BullMQ pipelines processing 12K+ jobs/day with retries and concurrency controls, sustaining p95 latency under 1.8s for real-time document automation. Implemented multi-tenant RBAC across 6+ microservices with IAM authorization and tenant-scoped access, reducing unauthorized access incidents and simplifying permission audits.

NestJS Redis BullMQ Kubernetes Helm OpenTelemetry Sentry Authentik OIDC LangGraph Multi Agent Systems OpenAI Anthropic Vertex AI Qdrant RAG MCP

Backend Engineer Intern

ShyftLabs India

Jun 2025 - Jun 2026 (1 year)

Productionized Kubernetes/Helm services with OpenTelemetry, Sentry, and Authentik/OIDC; improved deploy reliability and cut mean incident triage time via structured tracing. Designed a LangGraph multi-agent platform with 10+ agents for routing, planning, memory, and streaming; raised task completion rate by 28% on internal eval suites.