Contributed to established projects:
- JUnit5 - Java testing framework
- DBeaver - Universal database tool
- Google Blockly - Visual programming editor
- AsciiDoctor - Documentation tools
Local document Q&A using Llama 3.1, ChromaDB, and sentence-transformers. Complete RAG pipeline with Docker deployment.
Tech: Python | Ollama | ChromaDB | PyMuPDF
Highlights:
- Zero API costs with local LLM inference
- Configurable chunking strategies
- Retry logic for production reliability
Terms of Service RAG. Complete pipeline with Docker deployment and local.
Tech: Python | Ollama | ChromaDB | PyMuPDF
Highlights:
- Similar to RAG system but with an interface
Natural language to SQL converter with automatic error recovery. Features intelligent visualization and multi-attempt retry logic.
Tech: Python | Streamlit | Ollama | DuckDB | Plotly | Pydantic
Highlights:
- Self-healing queries (3-attempt retry with error analysis)
- Structured LLM outputs validated by Pydantic
- Auto-generated charts from query results
- Interactive Streamlit UI with query history
Java implementation of the ray tracer
Tech: Java
Highlights:
- Based on "Ray Tracing in a weekend"
Multi-season playoff outcome prediction using Elo ratings and ensemble methods. 72% accuracy on test set.
Tech: Python | scikit-learn | XGBoost | pandas | NumPy
Highlights:
- Custom Elo rating implementation
- 40+ engineered features
- SHAP explainability analysis
- Handles imbalanced playoff datasets
Competition entries covering classification, regression, and deep learning. Includes CNN implementations and feature importance studies.
Tech: TensorFlow | Keras | XGBoost | SHAP | scikit-learn
Highlights:
- 98.5% accuracy on MNIST (CNN)
- Imbalanced classification with SMOTE
- Feature engineering from domain knowledge
Languages & Frameworks
Java (Primary)
Python (Proficient)
C++ (Intermediate)
C# (Unity/Game Dev)
ML/AI Technologies
- LLMs: Ollama, LangChain, Hugging Face Transformers
- Frameworks: Spring Boot, PyTorch, TensorFlow/Keras, scikit-learn
- Vector DBs: ChromaDB, Weaviate, Pinecone
- MLOps: Docker, Docker Compose, Git
- Data: pandas, NumPy, DuckDB, Jupyter
- Visualization: Plotly, Streamlit, matplotlib, seaborn
Software Engineering
- Tools: Git, Docker, Linux, VS Code
- Testing: pytest, unittest, CI/CD basics
- APIs: FastAPI, Flask, REST architecture
- Databases: PostgreSQL, MongoDB, SQLite, DuckDB
LinkedIn: Connect with me