Mitul Chavda

Full-Stack Software Engineer

Full-Stack Engineer with 2.6 years of experience at Neome.ai, where I scaled core products from 0 to 1,000+ active users. Specialized in designing and building production-grade RAG pipelines, multi-agent systems, and high-concurrency real-time applications using TypeScript, Python, LangGraph, and modern LLM tooling. Delivered measurable efficiency gains through intelligent automation.

My Skills

React
Next.js
TypeScript
HTML5
CSS3
Tailwind CSS
Framer Motion
React
Next.js
TypeScript
HTML5
CSS3
Tailwind CSS
Framer Motion
Node.js
Express.js
PostgreSQL
Redis
Prisma
Java
REST APIs
Node.js
Express.js
PostgreSQL
Redis
Prisma
Java
REST APIs
WebSockets
Server-Sent Events
Caching
RBAC
Performance
WebSockets
Server-Sent Events
Caching
RBAC
Performance
Git
GitHub
Docker
AWS
Vercel
CI/CD
Git
GitHub
Docker
AWS
Vercel
CI/CD

Experience

My professional journey and key experiences

Software Engineer

Neomenta Pvt. Ltd.

Aug 2023 – Present
  • Architected scalable full-stack systems serving 2,000+ daily users — designed component-based React and React Native architecture with reusable patterns for state management and real-time data synchronization across web and mobile platforms
  • Built a visual query builder with autocomplete, validation, and real-time preview — cutting query creation time by 60% and boosting user productivity by 40%
  • Engineered an automated field mapping engine that resolves 50+ fields across forms and workflows using pattern matching and validation rules — eliminating ~90% of manual configuration effort and reducing setup errors to zero
  • Optimized Workflow Builder performance with diffing algorithms and WebSocket-based synchronization for drag-and-drop interactions — achieving 30% faster rendering through efficient state reconciliation
  • Developed real-time analytics infrastructure using WebSocket streams, server-sent events, and Redis caching — delivering sub-second data updates to 2,000+ daily users at 99.9% uptime
  • Built offline-capable React Native workflows and a shared component library spanning 3+ modules, accelerating cross-platform delivery
React
TypeScript
React Native
WebSockets
Redis

Frontend Developer Intern

TatvaSoft Pvt. Ltd.

Jan 2023 – Jun 2023
  • Led team of 3 to build modular LMS
  • Implemented JWT authentication flows
  • Designed UI/UX prototypes (1st Prize internally)
React
JWT
UI/UX Design

My Projects

SentinelInvoice

Autonomous AI invoice auditing engine that parses PDF invoices via LlamaParse OCR, extracts structured fields with Gemini, validates line-item math, and checks historical vendor prices (±5%). Flags anomalies to a Tavily-powered ReAct enrichment agent that computes a trust score (0–100); low-trust invoices pause at a LangGraph human-in-the-loop checkpoint before finalizing.

LangGraph
LangChain
Google Gemini
Groq
LlamaParse
Tavily
PostgreSQL
Express.js
React

Code-RAG

Production-grade codebase Q&A system that ingests any GitHub repo, parses it with tree-sitter AST, and stores a Neo4j knowledge graph of files, classes, and functions. Answers questions via hybrid RRF retrieval (vector + BM25 + graph traversal) fed into a CRAG corrective loop — grades each retrieved doc, rewrites the query if quality is low (max 3 iterations), then checks the answer for hallucinations before returning.

LangGraph
LangChain
Neo4j
Groq LLaMA-3.3
FastAPI
Next.js
tree-sitter
Python

FlashCommerce

High-traffic flash sale backend built to handle thousands of concurrent buyers without overselling. Uses Redis Lua atomic scripts to reserve inventory in microseconds, 15-minute TTL checkout reservations that auto-release on timeout, and two-layer storage (Redis for real-time operations, PostgreSQL for persistent orders). Includes JWT auth, Stripe payment integration, Swagger docs, and a Jest test suite.

Node.js
Fastify
TypeScript
Redis
PostgreSQL
Stripe
JWT
Docker
Jest

Satellite Surveillance

Real-time 3D satellite tracking app that visualises thousands of active satellites over an interactive Cesium.js globe. Streams live TLE orbital data from CelesTrak, classifies satellites by type (ISS, Starlink, GPS, Debris), and lets users click any satellite to fly to it and render its predicted orbit path. Backend caches TLE data in Redis for 12 hours to avoid rate-limiting.

React 18
TypeScript
Cesium.js
Node.js
Fastify
Redis
satellite.js

Let's Connect

Get In Touch

Email

mitulchavda100@gmail.com

Phone

+91 63540 34566

Location

Ahmedabad, Gujarat, India

Socials . . .