ML Engineer (Aspiring) · Full-Stack Developer · AI Systems & Deployment
Building AI-powered products end-to-end: data → modeling → evaluation → deployment
- Applied ML / Deep Learning: CNNs, transfer learning (MobileNetV2, ResNet, timm), image classification, model evaluation & ablations
- NLP & LLMs: Prompt engineering, structured generation, root cause analysis with LLMs
- MLOps & Deployment: Model serving via FastAPI/Flask, Docker, Hugging Face Spaces, Vercel
- Full-Stack AI Products: React/Flutter frontends + Python ML backends — shipping complete systems, not just notebooks
End-to-end AI healthcare platform — deep learning for dental disease detection from X-rays and photos
| Aspect | Detail |
|---|---|
| Problem | Automate dental disease screening to improve accessibility of dental care |
| Models | Custom CNNs trained with PyTorch + timm — separate models for normal dental images and X-rays |
| Architecture | Multi-head classification with confidence scores + personalized recommendations |
| Deployment | FastAPI backend deployed on Hugging Face Spaces (Docker), React web app + Flutter mobile app |
| Extra features | AI chatbot for dental queries, JWT auth, clinic locator via Geopify API |
| Tech | PyTorch timm FastAPI React/TypeScript Flutter Docker SQLite |
⭐ 4 stars · 5 forks · Capstone project (with Het Patel)
Deep learning for plant disease detection + ML-based irrigation scheduling
| Aspect | Detail |
|---|---|
| Disease Detection | MobileNetV2 with transfer learning → ~90% accuracy on PlantVillage dataset (38 classes) |
| Irrigation Model | Logistic Regression on tabular sensor data (temp, humidity, rainfall, soil moisture) → ~85% accuracy |
| Model Selection | Evaluated ResNet50 (95% acc, too heavy), VGG16, Random Forest, SVM, Naive Bayes — chose MobileNetV2 for accuracy/latency tradeoff |
| Deployment | Flask API backend + React/Vite/TypeScript frontend, deployed on Vercel |
| Tech | TensorFlow/Keras scikit-learn Flask React/TypeScript Vite |
⭐ 12 stars · 5 forks · Live in production · Team of 3
3. BugReport-AI
AI-powered platform for generating structured bug reports and root cause analysis
| Aspect | Detail |
|---|---|
| Problem | Automate the generation of well-structured bug reports from unstructured input; perform root cause analysis |
| Approach | LLM-powered pipeline for structured output generation — NLP + prompt engineering |
| Tech | Python (100%) · Currently in active development |
| Why it matters | Demonstrates NLP/LLM integration, structured generation, and developer-tooling AI |
ML/DL: PyTorch, TensorFlow/Keras, scikit-learn, timm, OpenCV, NumPy, Pandas
Languages: Python, TypeScript/JavaScript, Java, C++, Go, Rust, R, Dart
Backend & Serving: FastAPI, Flask, Django, Node.js/Express
Frontend: React, Next.js, Flutter, Tailwind CSS
Data/DB: PostgreSQL, MySQL, MongoDB, SQLite
DevOps/Cloud: Docker, AWS, GCP, Vercel, Hugging Face Spaces, Render
AI/ML Engineer · ML Intern · Data Scientist · Full-Stack ML Engineer
Interests: Computer Vision, NLP/LLMs, Healthcare AI, MLOps, ML Systems
I ship AI systems, not just train models.