Kenean Dita Meleta
Machine Learning & Backend Engineer
I build ML-driven systems and reliable backend APIs turning messy data and ambiguous requirements into production-ready solutions.
About
A quick snapshot of what I do and what I’m focused on.
I’m a Machine Learning & Backend Engineer focused on building end-to-end products: from data pipelines and model training to scalable APIs and deployment.
I like solving problems where accuracy, latency, and maintainability all matter especially when systems need to work reliably under real-world constraints.
- MLOps best practices
- Vector search + RAG patterns
- Distributed training
- Machine Learning Engineer
- Backend Engineer
- Data Analyst
Skills
Grouped for fast scanning no bars, just what I use.
Python
NumPy
Pandas
Scikit-learn
TensorFlow
PyTorch
Jupyter
Keras
Flask
DjangoREST APIs
PostgreSQL
MySQL
SQLAlchemy
Git & GitHub
Docker
Kubernetes
PostmanProjects
A selection of ML, backend, and CLI work-focused on outcomes and clarity.
Predict next-day crypto closing prices from today's market + technical indicators.
- Next-day price prediction for BTC, ETH, LTC, and XPR
- Uses 14 engineered features (RSI, MACD, MAs, volatility, returns)
- Auto-calculates missing indicators and supports sample input data
Result: A practical forecasting demo that turns raw OHLC into actionable next-day estimates.
Securely capture, organize, and retrieve personal memories (text, images, audio, video).
- Privacy-first storage with tagging, search, and timeline organization
- Media-friendly architecture using S3-style object storage and DynamoDB-style metadata
- Local dev environment via Docker + LocalStack
Result: A foundation for a personal knowledge/memory system with secure storage patterns.
Analyze and predict cryptocurrency market movements with a minimal ML stack.
- Interactive Streamlit app for exploration + prediction workflows
- Clear, minimal project structure focused on crypto signal iteration
Result: A lightweight crypto ML sandbox that ships reproducibly with Docker.
Boost terminal productivity with code generation, syntax highlighting, and interactive command prompts.
- CLI UX with rich formatting and interactive flows
- Designed to guide users from suggestion → execution safely
Result: A developer-tooling CLI that makes automation and iteration faster.
Deliver a modern, responsive e-commerce experience with clean UI and solid backend structure.
- Django-powered web platform with clean, responsive Tailwind UI
- Optimized for a smooth browsing and shopping flow across categories
Result: A production-style full-stack web app showcasing UX + backend fundamentals.
Predict student performance from input features with an end-to-end ML + web deployment workflow.
- Covers preprocessing, training, evaluation, and serving predictions
- Flask app provides real-time inference from user inputs
Result: A complete ML demo that connects model building to a usable web interface.
View your latest GitHub public activity directly in the terminal (commits, PRs, issues, events).
- Connects to the GitHub API and formats output for fast scanning
- Clean, readable terminal UX for recent activity summaries
Result: A small but useful CLI that turns noisy activity feeds into clarity.
Recognize handwritten Amharic fidel characters via a CNN model served through a Streamlit app.
- CNN model trained on a custom handwritten Amharic dataset
- Streamlit UI serves predictions for 34 root groups × 7 orders
Result: A complete deep learning demo combining training, inference, and a usable UI.
Contact
If you want to collaborate or chat about a role, reach out.
The fastest way to reach me is email. I usually reply within 24-48 hours.