Open to AI / ML roles · Based in India · --:-- IST
Multi-agent systems · RAG · LLM orchestration

Hi, I'm an
AI / ML Engineer

I design and ship production AI systems — agentic pipelines, retrieval-augmented LLMs, and computer-vision models — built to be grounded, observable, and actually deployable.

View Projects Contact Me Live demo · VERITAS
Top 30of 1,500+ · Zelestra × AWS
4-agentRAG, in production
92%PCB defect · CV
Chaitanya Chollangi
Chaitanya ChollangiAI / ML Engineer · India

Hello!

I'm Chaitanya — a Computer Science engineer from VIT Bhopal who loves turning AI ideas into real, working systems that hold up well outside the notebook.

I work across the AI stack, from LLM applications and retrieval pipelines to computer vision, and I care about building things that stay grounded, handle uncertainty, and run reliably in the real world. My keenest interest is in AI engineering and the agentic frontier — AI agents, voice agents, and the kind of software that can reason and act on its own — and I'm always chasing the next thing to build. Ranked top 30 of 1,500+ teams at the Zelestra × AWS ML Ascend Challenge 2025.

Python LangGraph LangChain FastAPI Docker AWS YOLO · CV SQL
Selected Work

Systems I've designed, built,
and shipped.

A few projects that show how I think about agents, retrieval, and reliability — from a full equity-research swarm to vision models on the factory floor.

Live
01 — Flagship

VERITAS · Equity Research Swarm

Multi-agent RAG · LangGraph · FastAPI + SSE

A production multi-agent system that reads SEC EDGAR filings and writes grounded equity-research memos. A team of LangGraph agents retrieves, drafts, and runs a self-checking critic loop before anything reaches the reader. It streams reasoning live over Server-Sent Events into a three.js / WebGL front end, and deliberately gives no buy/sell calls — a built-in safety stance. Hardened for public traffic with a rate limiter and result cache, and runs lexical-only retrieval to fit free-tier memory limits.

LangGraphRAGFastAPISSE streamingthree.jsSEC EDGAR
Live demo GitHub
02

Circuit Guard Pro

Computer Vision · YOLO · OpenCV

An automated PCB defect-detection system using a YOLO-based model to flag faults from board imagery. Reached 92% detection accuracy and cut manual inspection time by roughly 80% — turning a slow visual QA step into a near-instant one.

YOLOOpenCVPythonCV
GitHub
03

Plant Disease Prediction

Computer Vision · Deep Learning

A leaf-image classifier that identifies crop diseases from a single photo, built to give growers a fast, accessible first read on plant health from a deep CNN trained on labelled disease imagery.

PythonCNNImage classification
GitHub
04 — Internship · Apptad Technologies

Multilingual OCR & Translation

FastAPI · EasyOCR · Llama 3.2

A pipeline that extracts text from images across multiple languages and translates it end-to-end, wrapping EasyOCR with a Llama 3.2 translation stage behind a FastAPI service. Built during my AI/ML internship at Apptad Technologies (Feb–Apr 2026).

FastAPIEasyOCRLlama 3.2NLP
Internship project
How I build

A grounded, self-checking way to ship AI.

Agentic systems fail quietly. My process is built to catch that — every stage exists to keep the output honest, traceable, and safe to deploy.

01

Understand

Pin down the real task, the data, and the failure modes that matter before a single agent is written. Constraints first — memory, latency, cost.

02

Architect

Choose the orchestration that fits — LangGraph over heavier frameworks when control matters — and design retrieval so every answer traces back to a source.

03

Build

Ship a typed FastAPI backend with streaming, caching, and rate limits. Make the system observable so I can see what each agent actually did.

04

Validate

Add a critic loop that checks the model against its own evidence, and refuse to overstep — no confident answers the data can't support.

grounded & ready to ship!
Toolkit

The stack I build with.

From agent orchestration down to the model and the metal — the tools I reach for to ship grounded, production-ready AI.

01

Languages & Core

PythonC++SQLJavaScriptReact JSHTML / CSS
02

Agentic AI & LLMs

LangGraphLangChainRAGMulti-agent systemsFastAPISSE streamingChromaDBLlama 3.2Prompt engineering
03

Machine Learning & CV

PyTorchYOLOOpenCVCNNsXGBoostLightGBMscikit-learnEasyOCR
04

Tooling & Infra

DockerAWSGitthree.jsPostmanSeleniumPlaywrightAI IDEs
Contact
Let's build something
chollangichaitanya04@gmail.com github.com/Riishii21 LinkedIn