Looks like you have JavaScript disabled. For the full Twine experience, you will need to re-enable it.

I’m an AI engineer building production LLM systems, RAG infrastructure, and agentic workflows. Currently shipping AI at Master’s Union and solo-building Tanishi, an autonomous agent with multi-model routing, MCP integration, long-term memory, and self-correction. I focus on multi-agent systems, evaluation, and reliable deployment under real cost and latency limits.…

Jyotiraditya Mishra

AI Engineer, Full Stack Developer, AI Developer





I’m an AI engineer building production LLM systems, RAG infrastructure, and agentic workflows. Currently shipping AI at Master’s Union and solo-building Tanishi, an autonomous agent with multi-model routing, MCP integration, long-term memory, and self-correction. I focus on multi-agent systems, evaluation, and reliable deployment under real cost and latency limits.…

Available to hire

Gurugram, India

I’m an AI engineer building production LLM systems, RAG infrastructure, and agentic workflows.

Currently shipping AI at Master’s Union and solo-building Tanishi, an autonomous agent with multi-model routing, MCP integration, long-term memory, and self-correction. I focus on multi-agent systems, evaluation, and reliable deployment under real cost and latency limits.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Intermediate

Intermediate

Intermediate

Intermediate

Intermediate

Intermediate

Intermediate

Intermediate

Language

English

Fluent

Work Experience

AI / Software Engineer at Master’s Union

June 1, 2025 - Present

Architected role-aware multi-tenant chat with hybrid LLM orchestration routing between OpenAI Chat Completions and Assistants API, including thread lifecycle management, dynamic instruction injection, and provider fallback. Engineered SSE token-streaming pipeline with graceful disconnects, ownership validation, per-user rate limiting, and fault-tolerant message persistence under concurrent load. Designed normalized session/project/message schema with archival, regeneration, and token-usage instrumentation; shipped admin governance layer enabling non-engineers to configure assistants without deploys. Architected additive RAG pipeline on existing Next.js + PostgreSQL stack using pgvector; sub-50ms semantic search over 10K+ profiles with hybrid retrieval and Redis semantic caching projected to cut LLM spend 30-50%.