Nikunj Khitha
Shipping GenAI Platforms
I build full-stack GenAI platforms with TypeScript/Node.js, Go/Gin, Java/Spring, and Python/FastAPI - from GraphRAG and AI gateways to backend APIs and workflow automation.
About
Nikunj
I build full-stack GenAI platforms - retrieval, APIs, automation, observability, and product UX.
I build full-stack GenAI platforms that improve retrieval quality, lower model cost, and eliminate manual work. My work spans GraphRAG and LightRAG pipelines, TypeScript/Node.js and Go/Gin services, Java/Spring and Python/FastAPI backends, AI gateways and observability, and the automation systems that make enterprise AI reliable at scale. I enjoy owning the stack end to end - from data and APIs to infrastructure and polished product experiences.
Architected GenAI platforms for 200+ customers, including a GraphRAG system unifying 500K+ entities across 5+ enterprise systems.
Cut indexing cost by 50%, saving $15K+ annually, while improving retrieval accuracy by 40%.
Built automations across support, QA, HR, and marketing that eliminate 200+ manual hours every month.
Led documentation automation for 250+ integrations, reducing update latency by 99% from 72 hours to 45 minutes.
Professional
Experience
From public-sector software to enterprise GenAI platforms, I focus on measurable gains in retrieval quality, cost, uptime, and team productivity.
And a new adventure ahead
Software Development Intern
Central Electricity Authority, Government of IndiaExperienceMay 2023 - July 2023- Enhanced a national renewable energy dashboard for 150+ power stations by integrating National Power Portal data in Java, improving accuracy by 30%.
- Built a secure PHP and MySQL file management system with role-based access control, improving retrieval efficiency by 25% across 5,000+ files.
- Developed a full-stack MERN conference room booking system that cut booking time by 60% and reduced scheduling errors by 40%.
SDE Intern (Backend/AI)
Xansr Media (Aiko)ExperienceJun 2024 – Dec 2024- Built Node.js and FastAPI microservices with test-driven development, achieving 100% test coverage and improving API performance by 40%.
- Designed Docker and GitHub Actions CI/CD pipelines that reduced deployment time by 42%.
- Engineered Fantasy GPT with RAG and LangGraph to achieve 98% complex-query resolution, and built AIKO for personalized sports highlights with 96% commentary accuracy.
- Created scalable MSSQL and Azure ETL pipelines that maintained 100% data accuracy for AI products.
AI Automation Intern
ArmorCodeExperienceJan 2025 - Nov 2025- Built backend capabilities for the core platform agent with Java and Spring Boot, including new APIs, prompt refinements, and AWS S3 vector knowledge base workflows.
- Led AI-driven code-to-documentation automation with CrewAI and MCP servers for 250+ security integrations, reducing update latency by 99% from 72 hours to 45 minutes.
- Created an OpenAI-compatible proxy for Gemini and Claude, and deployed LiteLLM to monitor 15+ AI APIs with centralized cost visibility.
Associate Engineer (Platform & GenAI)
ArmorCodeExperienceJan 2025 – Present- Architected enterprise GenAI platforms for 200+ customers, including a Neo4j + PGVector GraphRAG system unifying 500,000+ entities across 5+ systems with 40% better retrieval accuracy.
- Orchestrated GraphRAG and LightRAG ETL pipelines that cut LLM indexing costs by 50% and saved $15,000+ annually.
- Expanded the core Java and Spring Boot backend with 10+ REST APIs serving 5,000+ daily requests at 99.8% uptime.
- Built automation suites across support, QA, HR, and marketing with n8n, Java microservices, and Python, eliminating 200+ manual hours each month.
And a new adventure ahead
Featured Work
Production-grade systems, AI platforms, and scalable applications built to solve complex problems.
CodeNex: AI Builder
AI-assisted full-stack app builder inspired by Lovable/v0, built as a complete product platform with a microservices Java backend and connected frontend workflows for auth, dashboards, usage views, and project building.
CodeNex AI API Proxy
Unified AI gateway bridging local AI tools to the web. Single API interface to multiple LLM providers with protocol translation, provider pooling, health-tracked failover, and streaming.
Serenify
Open-source mental wellness app combining empathetic AI chat (Gemini 2.5 Flash), mood tracking, journaling, guided wellness sessions, and personal analytics with privacy-first design.
CodeNex Images
AI-powered image generation and editing platform using Google Gemini models, with Auth0 authentication and a polished editing workspace.
Resume Fit — CodeNex
AI-powered resume analysis & optimization tool with ATS score checking, multi-pillar performance scoring, refinement workflows, and keyword extraction.
LLaMa MCP Streamlit
Interactive AI assistant using Streamlit, NVIDIA NIM's LLaMA 3.3 70B, and Model Context Protocol (MCP) for real-time external tool execution.
Technical
Arsenal
A production-tested stack shaped by TypeScript/Node.js, Go/Gin, GraphRAG, AI automation, backend platforms, CI/CD, and polished full-stack product delivery.
Generative AI & ML
20 technologies
Full Stack Development
19 technologies
Databases & Data
13 technologies
DevOps & Infrastructure
11 technologies
Dev Tools & Platforms
14 technologies
Talk to my
AI Twin
Curious about my work? My AI twin can walk you through my projects, architecture decisions, and experience — just like I would.
Get in Touch
Open to new opportunities, collaborations, and interesting conversations.
Let's Build Something
Whether you're looking to hire, collaborate on an open-source project, or just want to chat about GenAI - I'm all ears.