nilotpal = {
"name" : "Nilotpal Dhar",
"role" : ["Deep learning", "Machine learning Engineer", "Open Source Developer"],
"education" : "B.Tech CSBS — Academy of Technology, West Bengal (2023–2027)",
"kaggle" : "Expert 🏆 — competition medals & high-ranking finishes",
"pypi" : "datadiagnose — zero-dependency ML dataset diagnosis library",
"location" : "West Bengal, India 🇮🇳",
"stack" : ["Python", "PyTorch", "Scikit-Learn", "FastAPI", "SQL"],
"live_apps" : "5 deployed ML projects — each with a public URL & live backend",
"available" : True,
"looking_for" : ["Deep Learning & ML Engineer", "Freelance AI Consulting"],
"contact" : "dharnilotpal31@gmail.com",
}I'm a Deep learning and ML Engineer currently in my final year student of Computer Science
I hold Kaggle Expert status earned through competition medals and real-world dataset finishes. I've built and deployed 5 live ML applications covering fraud detection, medical imaging, rent prediction, churn analysis, and NLP recommendation — every single one has a live URL and a FastAPI backend on Render, not just a notebook.
I'm also the author of DataDiagnose — a Python library published on PyPI with zero external dependencies. It auto-diagnoses ML datasets, scores dataset health from 0–100, and recommends the right model type before training begins. Built from scratch using only the Python standard library. 140-test suite. MIT licensed.
My standard: Does it solve a real problem? Does it actually work? If yes, I ship it
Every project below has a live URL — FastAPI backend on Render, frontend on Vercel or GitHub Pages.
| # | Project | What it does | Tech | Links |
|---|---|---|---|---|
| 1 | 🔬 DermSight PRO | ResNet-50 skin lesion classifier · 85%+ acc · 7 ISIC classes · HAM10000 dataset | PyTorch · FastAPI · ResNet-50 |   |
| 2 | 🛡️ FraudGuard AI | Real-time credit card fraud detection microservice with live dashboard | LightGBM · FastAPI · Vercel | |
| 3 | 🏠 RentIQ | AI rent forecasting across 6 Indian metro cities using tuned XGBoost | XGBoost · FastAPI · GitHub Pages | |
| 4 | 📡 RevenueShield ML | Telecom churn predictor with real-time risk scoring web app | Random Forest · FastAPI | |
| 5 | 🎬 Movie Recommender | NLP content-based recommender · Streamlit dashboard · 5,000-movie corpus | NLP · Scikit-Learn · Streamlit | |
pip install datadiagnosefrom datadiagnose import DataDiagnose
import pandas as pd
report = DataDiagnose(pd.read_csv("dataset.csv")).diagnose()
# → health score 0–100, detected issues, model type recommendationsDetects: missing values · duplicates · class imbalance · data leakage · high cardinality · skewed features · outliers · constant columns
| Issuer | Certificate | Status |
|---|---|---|
| Oracle | Cloud Infrastructure Data Science Professional |  |
| Oracle | Cloud Infrastructure AI Foundations Associate | |
| LinkedIn & Microsoft | Career Essentials in Data Analysis |  |
| IBM — Cognitive Class | Python 101 for Data Science |  |
| Udemy | NumPy · SciPy · Matplotlib · Pandas |  |
Languages
ML / Deep Learning / Data Science
Deployment & APIs
Databases & Tools
From raw data to a deployed, working solution — I'm ready.