I'm a Data Scientist who enjoys building software just as much as building models.
Most of my work sits somewhere between data science and backend engineering—building APIs, deploying machine learning models, integrating LLMs, and turning ideas into production software. I like owning projects end-to-end, from figuring out the statistical approach to shipping something customers can actually use.
A retrieval application for analyzing historical EIA energy forecasts. It combines deterministic forecast analysis with hybrid retrieval (BM25 + dense retrieval) to answer questions using cited source material instead of relying on a general-purpose LLM.
Things I explored
- Hybrid retrieval
- LangGraph workflows
- Retrieval evaluation
- FastAPI
- Docker
- Automated testing
Tech: Python, FastAPI, LangGraph, BM25, BGE Embeddings, Pydantic, Docker
A production-oriented machine learning API focused on building and deploying reliable forecasting models rather than maximizing leaderboard accuracy.
Highlights
- Chronological validation
- Leakage prevention
- Model versioning
- Prediction intervals
- Drift diagnostics
- CI/CD
Tech: Python, FastAPI, XGBoost, scikit-learn, Docker
An API that combines deterministic dataset profiling with structured LLM outputs to generate consistent data briefings.
Highlights
- OpenAI Agents SDK
- Structured outputs
- Pydantic models
- Typed API design
- Behavioral evaluation
- Docker
Tech: Python, Quart, OpenAI Agents SDK, Pydantic, Docker
Languages
- Python
- SQL
- JavaScript
AI & Machine Learning
- Machine Learning
- LLM Applications
- RAG
- LangGraph
- OpenAI APIs
- OpenAI Agents SDK
- XGBoost
- scikit-learn
- TensorFlow / Keras
Backend
- FastAPI
- Quart
- Pydantic
- Docker
Cloud & Data
- Azure
- Azure Container Apps
- Azure Functions
- PostgreSQL
- SQL Server
- Snowflake
- Azure Cosmos DB
I'm currently a Data Scientist at ShareBuilders, where I build forecasting systems, production APIs, and AI features for a SaaS platform. Lately I've been spending a lot of time learning more about retrieval systems, agent workflows, and modern AI engineering.
If you're interested in production ML, AI applications, or backend engineering, feel free to connect.
💼 LinkedIn: https://www.linkedin.com/in/kyle-dinges