rag.retrievalExpert
RAG & Retrieval
- ›Hybrid search & reranking
- ›Pinecone · pgvector · FAISS
- ›Chunking strategy & RAG evals
- ›Citation-grounded answers
AI/ML Engineer
Hi, my name is
HemangBaldha.
I take RAG systems to production.
AI engineer with 3 years of experience building intelligent systems end-to-end. RAG pipelines that answer with citations, LLM agents that do real work, and voice AI that listens and talks back. From first prototype to production traffic, I own the whole journey.
RAG✦LLM Agents✦Whisper / Voice AI✦Fine-tuning · LoRA✦Vector Databases✦Computer Vision✦LangChain / LangGraph✦PyTorch✦Hugging Face✦FastAPI✦Next.js✦AWS✦
01who I am
About Me
Hello! I'm Hemangkumar Mathurbhai Baldha, an AI engineer with 3 years of experience building intelligent systems that make it to production - not just notebooks. I care about the unglamorous parts that make AI actually work: retrieval quality, latency budgets, evals, and clean data.
My day-to-day is production RAG systems, LLM integrations and agentic workflows, and voice & audio pipelines - rounded out by fine-tuning and computer-vision work. Lately that means multi-tenant RAG platforms, an NL-SQL engine that queries 20+ databases, and LoRA-fine-tuned vision-language models watching real warehouses.
I enjoy leading technical projects and bridging AI research with applications people actually use.
Technologies I've been working with recently:
- Python
- PyTorch
- LangChain / LangGraph
- Hugging Face
- Whisper / TTS
- OpenCV
- Pinecone / pgvector
- FastAPI
- Next.js / React
- TypeScript
- PostgreSQL
- AWS / Docker
3+years building AI systems
2xhackathon wins - 1st place
500+students taught
02the toolkit
What I Actually Do
The honest version of a skills chart - where I'm deep from 3 years of building AI systems, and where I'm still leveling up.
llm.integrationExpert
LLM Integration & Agents
- ›OpenAI · Claude · open-weights
- ›Function calling & tool use
- ›LangChain · LangGraph workflows
- ›Streaming, guardrails, cost tuning
audio.voiceExpert
Voice & Audio AI
- ›ASR / TTS pipelines (Whisper+)
- ›Audio processing for voice models
- ›Real-time streaming audio
- ›VAD & speaker diarization
model.finetuneProficient
Fine-tuning
- ›LoRA · QLoRA · PEFT
- ›Dataset curation & cleaning
- ›Instruction tuning
- ›Evaluation loops
visionProficient
Computer Vision
- ›Detection & classification
- ›OCR & document AI
- ›OpenCV pipelines
- ›Image preprocessing
fullstack.servingExpert
Full-Stack & Serving
- ›FastAPI · Node services
- ›Next.js · React frontends
- ›Postgres + pgvector
- ›Docker · AWS deploys
03where I've been
Experience
AI/ML Engineer @ Atlasml.ai
April 2026 - Presentwww.atlasml.ai
Architecting a scalable multi-tenant AI platform with Retrieval-Augmented Generation (RAG), vector databases, agentic orchestration, memory systems, and tool integrations for intelligent workflows.
Built a natural language to SQL (NL-SQL) engine capable of querying 20+ heterogeneous databases and thousands of tables with schema-aware reasoning and optimized query generation.
Fine-tuned a Vision Language Model (VLM) using LoRA for automated cement warehouse inventory monitoring, enabling stock tracking, anomaly detection, and theft prevention.
AI/ML Engineer @ Cybercom Creation
Jan 2026 - March 2026www.cybercom.co.in
Developed and deployed AI solutions for the live Gujarat real estate platform (RERA).
Built production-grade intelligent systems using Retrieval Augmented Generation (RAG) architectures.
Designed Agentic AI workflows integrating LLMs, vector databases, and real-time data pipelines.
Implemented full lifecycle ML deployment including model development, evaluation, and production integration.
Software Engineer @ Sampaarsh Technologies
August 2024 - October 2025www.sampaarsh.com
Developed scalable web applications using React (Vite), Node.js, and MySQL with Agile/Scrum methodologies.
Worked on diamond cutting & testing tools to ensure high-quality releases using C++ frameworks like Qt, deployed with AWS on Linux environments.
Teaching Assistant (during undergraduation) @ Darshan University
December 2023 - September 2025www.darshan.ac.in
During semesters 4 and 5, acted as a Teaching Assistant for the 'Web Development' course, supporting students with HTML, CSS, JavaScript, React, Node.js, Express, and MongoDB.
In semesters 6 and 7, instructed undergraduate students in graph theory, data structures and algorithms, computer networks and machine learning, providing guidance on practical applications and problem-solving techniques.
Graduated more than 500 students to date.
04what I've built
Some Things I've Built
Enterprise AI Platform
Multi-Tenant AI Analytics & Agentic Platform
Built an enterprise-grade multi-agent AI platform that orchestrates agents for NL-to-SQL generation, SQL validation, query self-healing, retrieval, authorization, and AI-powered summarization with secure multi-tenant isolation. Implemented semantic schema indexing using OpenAI Embeddings, ChromaDB, and Redis caching, enabling dynamic onboarding of PostgreSQL, MySQL, MSSQL, Snowflake and many more databases for natural-language analytics.
- FastAPI
- CrewAI
- gRPC
- Docker
- PostgreSQL
- MySQL
- MSSQL
- Snowflake
- OpenAI
- ChromaDB
- Redis
NPM Package
Aliveicon - Animated Icons Library
An animated icon library with 1200+ ready-to-use icons for React, Vite, Next.js, or any standard web project - install it and start using icons instantly, no setup required. Built with Framer Motion and Tailwind CSS, it ships smooth, customizable animations with options for size, color, animation effects, hover behavior, loops, and more.
- TypeScript
- NPM
- Framer Motion
- Tailwind CSS
Skill Learning Platform
CyberXplained
An interactive platform that makes learning cybersecurity engaging - live command execution, attack simulations, a real-time encryption demo, a security toolbox, and a Stack Overflow-style forum.
- React
- Tailwind CSS
- TypeScript
- Framer Motion
Hackathon Project · 1st Place
Rajwadi Poshak
An AI-powered fashion e-commerce platform with smart recommendations and personalized styling - took 1st place at a hackathon, earning recognition, prizes, and internship opportunities.
- Next.js
- TypeScript
- Framer Motion
- Radix
Next.js Project
The DevEchelons
A competitive programming platform with a custom code editor where students solve problems and collaborate on challenges - in the spirit of LeetCode, Codeforces, and CodeChef.
- Next.js
- Tailwind CSS
- TypeScript
- MongoDB
First Hackathon · 1st Place
The Academia Hub
A platform where students learn skills and get certified, make friends, and connect over chat or video calls. Built at a national-level hackathon - secured 1st rank, my first hackathon.
- Angular
- TypeScript
- Socket.io [WebRTC]
05what's next
Let's build something intelligent together.
I'm always open to new opportunities, and my inbox is always open. Whether you have a question, a role in mind, or just want to say hi - I'll do my best to get back to you.
or write to hemang9705@gmail.com