AI/ML Developer

I build AI
systems that
actually ship.

RAG pipelines. Computer vision. LLM applications.
From model to production - every project ends with a live, usable app. Based in Nallasopara, Maharashtra.

3+
live deployments
98%
peak accuracy
5+
AI certifications
01 - PROJECTS

Things I've
built & shipped.

001 - COMPUTER VISION
FreshHarvest AI
Automated fruit freshness inspection using 3-phase ResNet50 transfer learning. Classifies 8 fruits as fresh or spoiled across 16 classes in real-time. Solved a domain-shift problem by auditing dataset diversity, rebuilding training data, and boosting diversity score from ~20 to 53.7.
▲ ~98% test accuracy · 16 classes
PyTorch ResNet50 Transfer Learning Streamlit rembg Git LFS
Live demo ↗ GitHub ↗
002 - RAG / LLM
MediInsight AI
Clinical decision-support system grounded in peer-reviewed literature. Ingests 300 PubMed abstracts via eutils API into ChromaDB, then answers clinician queries using Llama 3.3-70b (Groq API) - every response cites exact PMIDs. Zero hallucination without sources.
300 PubMed abstracts · Llama 3.3-70b
RAG ChromaDB Llama 3 Groq API TF-IDF Streamlit
Live demo ↗ GitHub ↗
003 - MULTI-CHAIN LLM
Intent-Based E-Commerce Chatbot
Semantic routing architecture that dispatches queries across three chains: RAG for FAQs, SQL agent for data queries, conversational chain for general chat. No irrelevant responses. Secure environment-based API key management for production deployment.
3-chain routing · Llama 3 · Groq
Semantic Routing RAG SQL Agent Llama 3 Groq API Streamlit
GitHub ↗
004 - ML ENGINEERING · CLIENT PROJECT
CodeX Beverage - Price Range Prediction
End-to-end ML solution delivered for a real F&B client at AtliQ Technologies. Cleaned 30,010 survey rows across 4 phases of EDA and feature engineering. Evaluated 6 classification models - LightGBM achieved 92.16% accuracy. Experiments tracked on MLFlow, model versions managed on DagsHub. Findings delivered via a 9-slide business presentation.
▲ 92.16% accuracy · 30,010 rows · Real client
LightGBM MLFlow DagsHub SQL Streamlit Python
Live demo ↗ GitHub ↗
02 - SKILLS

Tech stack.

DEEP LEARNING & VISION
PyTorch
Transfer Learning
ResNet50 / CNNs
LLM & GENERATIVE AI
RAG Pipelines
Prompt Engineering
ChromaDB / Vectors
Groq API / Llama 3
LANGUAGES & CORE
Python
SQL
Pandas / NumPy
MLOPS & DEPLOYMENT
Streamlit
MLFlow / DagsHub
Git / GitHub
FastAPI
03 - ABOUT

The person
behind the code.

I'm a fresher AI/ML developer from Nallasopara, Maharashtra. I completed my B.Sc. in Information Technology from D.G. Ruparel College, Mumbai in 2025.

My focus is on building things that work in production - not just Jupyter notebooks. Every project I build gets deployed as a live app. I care about the full pipeline: data → model → API → user interface.

I use AI tools like Claude and ChatGPT as force multipliers - to move faster, debug smarter, and write better documentation. But I own and understand every line that ships.

Currently looking for an AI/ML internship where I can work on real systems, learn from experienced engineers, and ship things that matter.

CERTIFICATIONS
Gen AI & Data Science Bootcamp 3.0
Codebasics
Apr 2026
Gen AI to Agentic AI
Codebasics
Mar 2026
Natural Language Processing
Codebasics
Jan 2026
Deep Learning: Beginner to Advanced
Codebasics
Dec 2025
Master Machine Learning for Data Science & AI
Codebasics
Aug 2025
04 - CONTACT

Let's build
something.

Building something interesting and need someone who ships?
Let's talk.

↗ Message on LinkedIn dipankarmane17@gmail.com