I build analytical tools across many domains, including:
Open source intelligence & policy analysis — NLP pipelines for unstructured government documents, cross-referencing defense budgets, policy documents, and program records to surface insights that aren't visible in any single source.
Dynamics & systems engineering — predictive modeling, simulation, and remaining useful life estimation for mechanical and aerospace systems. From turbofan degradation to SLAM-based robotics.
Business intelligence — translating messy operational data into structured, queryable systems with clear analytical outputs. Building the pipelines that turn documents and databases into decisions.
A full end-to-end pipeline that cross-references 28 years of DoD unclassified R-1 RDT&E budget documents with National Defense Strategies, National Security Strategies, and NDAA legislation — answering the question: did funding actually follow stated strategic priorities?
- Parses 28 fiscal years of PDF budget justification books into a structured SQLite database
- 4-stage NLP matching pipeline: PE number lookup → acronym index → fuzzy matching → semantic similarity
- Cross-references 12,000 policy document chunks against 2,100 program elements using SentenceTransformers
- Deployed as a Streamlit app via Docker on Railway
Stack: Python · SQLite · SentenceTransformers · RapidFuzz · Streamlit · Docker
| Project | Description | Stack |
|---|---|---|
| FireScout-SLAM | SLAM-based firefighting scout robot simulation | Python · ROS · Jupyter |
| Drone-Landing-Vision | Precision drone landing using YOLOv8 Nano on the TEKNOFEST dataset | Python · YOLOv8 · OpenCV |
| NFL Play Predictor | Next-play prediction from game state variables | Python · sklearn · Jupyter |
| NASA Turbofan RUL | Remaining useful life prediction using LSTM on the NASA CMAPSS dataset | Python · PyTorch · Jupyter |
Python PyTorch NLP Computer Vision Predictive Modeling Docker SQLite Streamlit PDF Parsing Data Engineering Tensorflow Data Analysis Data Engineering LLM RAG Neural Networks