Success storyAI & data
Context-aware SQL generation
Turning plain-English business questions into SQL people can actually trust — by giving a language model the schema, the vocabulary, and a reviewer.
- LLMs
- SQL
- Metadata
- Human review
Data & analytics · AI/ML · Data strategy
Hi, I'm Steven. I turn messy data into things people can actually use.
I'm a data and analytics leader in Virginia Beach — Director, Data & Analytics at Catalight, and founder of Fruitful Intelligence, my data science consultancy. I've spent my career turning payments, healthcare, tax, and public-sector data into models, platforms, and decisions, and teaching graduate data analytics along the way. This is where I write, experiment, and keep the side projects my GitHub keeps promising to finish.
Personal: email meBusiness & contracting: SoFruitful.com
📍 Virginia Beach, VA
Selected work
Things I've built
Success storyAI & data
ID validation
Cutting repetitive manual checks and inconsistent calls in document validation — with OCR, field-level checks, confidence signals, and a human for the hard cases.
- OCR
- Document AI
- Confidence scoring
- Human review
Open sourceAI & data
PyLytics (opens external site)
A desktop app for early-stage data modeling: load a CSV, explore distributions and correlations, plot, and train and compare scikit-learn models without writing a notebook.
- Python
- Tkinter
- scikit-learn
- pandas
ExperimentAI & data
Gradient boosting tuning experiments
Hands-on experiments in scaling and tuning gradient boosting classifiers on a public health-indicators dataset, written up as a blog post with code on GitHub.
- Machine learning
- Optimization
- Python
Recent writing
From the blog
5 min read
Beer
Copper Mosaic IPA: My Latest Homebrewing Adventure
A five-gallon Mosaic IPA experiment, an enthusiastic fermentation, and a beer that tasted suspiciously like a stout.
4 min read
Data Science · Development
Fine Tuning Justice: The Role of Pre-Trained LLMs in Enhancing Federal Investigations and Legal Procedures
Large language models (LLMs) represent the next generation of artificial intelligence applications, attaining widespread attention and adoption. These models demand substantial energy and resources for training. Consequently, there has been a shift towards developing pre-trained models like bidirectional encoder representations from transformers (BERT), utilizing millions of parameters from training texts to create general use models. An evolution of this approach involves fine-tuning models with domain-specific data, enhancing their utility in fields such as medicine, law, and science.
3 min read
Data Science
Optimizing Gradient Boosting Models
Gradient boosting classifier models are a powerful type of machine learning algorithm that outperform many other types of classifiers. In simplest terms, gradient boosting algorithms learn from the mistakes they make by optmizing on gradient descent. A gradient boosting model values the gradient descent, or the direction of the steepest increase of a function, to make adjustments so that the function can increase rapidly over each iteration. Gradient boosting models can be used for classfication or regression. This post will concern itself with gradient boosting classification. In particular, we will study and optimize three gradient boosting models, AdaBoost (sklearn), XGBoost, and Gradient Boosting Classifier (sklearn). We will study these classifiers with the Kaggle data set, Kaggle: Diabetes Health Indicators Dataset . In particular, we will use the binary data set. I will not be doing any cleansing or engineering of the data. I am strictly experimenting with the effects of the learning rate and number of estimators hyperparameters on gradient boosting model performance.
5 min read
Data Science
Traditional Methods vs AI Methods in Forecasting
Forecasting remains an indispensable tool for businesses to make informed decisions about future trends, demands, and opportunities. Accurate forecasts can lead to better resource allocation, inventory management, and strategic planning. When it comes to forecasting, data scientists have many tools at their disposal, ranging from traditional and time-tested methods to advanced artificial intelligence (AI) techniques. While AI methods have gained significant attention in recent years, traditional forecasting methods like Moving Average, Exponential Moving Average, and Linear Regression Forecasting still hold their ground in many real-world scenarios. As a data scientist with substantial experience in machine learning and big data analytics, I will shed light on why, in most cases, traditional forecasting methods remain preferable to their more advanced counterparts.
Off the clock
When I'm not wrangling data
I consider myself a generalist who's always chasing the next recreational coding project. Away from the keyboard, you'll usually find me outside or elbow-deep in a hobby. More about me →
- Kayaking & the Chesapeake Bay
- Hiking & backpacking
- Rock climbing
- 3D printing
- Hazy IPAs & homebrewing
- Sci-fi & fantasy novels
- Comics & action figures
- Pokémon (still)
Get in touch
Two doors, depending on why you're here
Personal
Just want to talk shop?
Questions about a post, a side project, a good IPA recommendation, or something you're building — my inbox is open.
Email mesteven.ray.purcell@gmail.com
Business & contracting
Have a project in mind?
Consulting, contracting, and other business inquiries go through Fruitful Intelligence LLC, my separate company site — not this personal inbox.
Visit SoFruitful.com ↗