Have challenging data? Do you need extra analytic expertise or just have questions for us? Please drop us a line.We're interested in learning more about your organization and the potential for data in helping advance its objectives. There's no fee or commitment for an introductory, solution-consulting conversation; we'll just listen, learn and provide honest, objective thoughts on next steps.Contact with us via LinkedIn or GitHub or by submitting the form below.

• LinkedIn
• GitHub

Download our Brochure

Stephen Godfrey

⦿ Steve Godfrey is a technical product manager and data scientist. In those roles, he helps organizations leverage data and analytics to improve their products and services and to support their missions.He frequently writes and speaks about tools and methods to manage, analyze and model datasets as well as applying data analytics in real-world use cases.His professional career has been in financial and technology services working for investment management, banking and financial technology firms including PayPal in data science and Wells Fargo and Bank of America in currency trading, global payments and risk-management research. In addition, his volunteer work includes evaluating statistical techniques, constructing machine-learning pipelines and software consulting for organizations such as Statistics Without Borders and the Technology Accelerator Fund.He holds an MBA and BS Mathematics/Applied Economics from UCLA and completed a data-science bootcamp. Steve lives in Palo Alto, CA with his wife, has two grown children and enjoys travel and outdoor activities and working with wood and in the garden.

• LinkedIn
• GitHub
• Email

• Projects
• Writings
• Home

Nathan Moroney

⦿ Nathan Moroney is an experienced research engineer. He has worked across organizations to invent, develop and ship a range of technologies. Nathan has written over 90 papers and patents with over 1,000 combined citations. He has been collecting and analyzing web-based data sets since before the term crowdsourcing was coined.He started as a research consultant at the RIT Research Corporation and then joined Hewlett-Packard Español (Barcelona) as a writing systems engineer. Nathan then transferred to HP Labs (Palo Alto) where he contributed to imaging, data, 3D, sensor and machine learning projects. His accomplishments as a principal scientist were recognized with the awarding of Fellow status at the Society for Imaging Science and Technology.Nathan has a BS degree from Jefferson University and Master of Sciences degree from the Rochester Institute of Technology. Both of his degrees are in the field of color science and his masters research topic was color space selection for JPEG image compression. He and his wife are Palo Alto residents with two grown children. When he cooks, it is often with one (or more) skillets and his current deep reading project is about rivers of the world.

• LinkedIn
• GitHub
• Email

• Projects
• Writings
• Home

California Community College Policy Assistant

⦿ Overview: An AI tool providing in-depth information to stakeholders in policy decisions affecting California Community Colleges⦿ Our contribution: The CCC Policy Assistant chatbot is an artificial intelligence (AI) agent trained on policy topics related to California's Community Colleges.The bot's target audience are stakeholders who would like to participate in community college decision making and would benefit from curated and detailed information related to community colleges. Some examples might include board members, administrators, staff, students, community activists or legislators.Our goal in creating it is to demonstrate how advocates can use technology to help further their missions.Some sample questions include:
- How many districts are there in the California community college system?
- What is the total enrollment of Foothill college?
- What college is designated a Center of Excellence in bioprocessing?
- How many California community colleges partner with the California Department of Corrections and Rehabilitation (CDCR) to provide in‑person courses?
- What are the responsibilities of the board members of a California community college?⦿ Resources:
  - Chatbot
  - GitHub

• Writings
• Team
• Home

Mental Health Web Application for Schools

⦿ Overview : A data-driven web application helping schools improve their mental health, trauma, social emotional learning and equity student supports.⦿ Our contribution : Using Python and cloud database tools, we helped build pipelines to collect, clean and enhance both publicly available and privately procured data. Data are stored in a data warehouse within a cloud data science environment and are available research, analysis and pre-processing before being passed to a front-end web application for presentation to school users. In addition to data analytics, the Numantic team provided product management expertise and assistance in designing early versions of the application.⦿ Resources:
  - RSSI

• Writings
• Team
• Home

Social Media Sentiment

⦿ Overview: A natural language processing (NLP) sentiment analysis tool pulling relevant Reddit posts and X (Twitter) tweets and assessing sentiment about homelessness in the Capital Region of British Columbia. This project was a collaboration between Statistics Without Borders and the Alliance To End Homelessness in the Capital. The work is being presented at the 2024 Joint Statistical Meetings (JSM).⦿ Our contribution: Using Python, we built a data-collection pipeline that employs application programming interfaces (API) from X (Twitter) and Reddit to collect tweet and post data and machine learning models to assess their relevance and sentiment. The pipeline runs on a Google Cloud instance and results are published to a public website.⦿ Resources:
  - GitHub

• Writings
• Team
• Home

OpenGazette - Patent Drawing Retrieval Tool

⦿ Overview: Thousands of new patent drawings are published each week by the US Patent Office – how can we better search this collection of images?⦿ Our contribution: We leverage image vector embeddings to create a database of weekly Patent Gazette drawings. This database is then queried using text descriptions, example images or categorical prompts. The embeddings and retrieval are performed using a Python toolbox running locally to achieve more stringent privacy requirements.⦿ Resources:
  - GitHub

• Writings
• Team
• Home

Stable Diffusion Color Analysis

⦿ Overview : How many color terms does a text to image generative AI learn through its training process?⦿ Our contribution : We developed batch mode prompt rendering using Python and proposed an initial automated analysis process. The analysis uses depth contours to estimate regions of interest and quantifies color differences with respect to a million-color term human corpus. We then use these results to estimate the range of color terms available for use.⦿ Resources:
  - Conference Proceedings Paper

• Writings
• Team
• Home

SWB & IMPACT INITIATIVES

⦿ Overview : What statistical methods should be employed to account for survey design effects?⦿ Our contribution : In this blog, we discuss a recent collaboration between Statistics Without Borders (SWB) and IMPACT Initiatives addressing survey design effects in the Multi-sectoral Needs Assessments (MSNA) conducted by IMPACT in countries experiencing humanitarian challenges and crises.⦿ Resources:
  - Blog

• Writings
• Team
• Home

AI Data Analysis

⦿ Overview : How can current AI tools be used by a data analyst?⦿ Our contribution : In this electronic book on AI Data Analysis, we explore using Large Language Models (LLM) to analyze datasets with the goal of developing some operating instructions or guidelines on how to effectively use these tools in the data-analysis task.⦿ Resources:
  - Blog

• Writings
• Team
• Home

Web-Scraping Pipeline

⦿ Overview : What pandemic-relief resources are available to small businesses?⦿ Our contribution : This video and paper cover a collaboration between Statistics Without Borders (SWB) and Client to Consultant Bridge (C2CB) to develop and deploy a multistage data pipeline to automatically curate a national list of small-business aid programs during the COVID-19 pandemic shutdown period. The pipeline's results were presented at a website and helped small business users efficiently research and find relevant aid programs.⦿ Resources:
  - Chance Journal Paper

• Writings
• Team
• Home

A TensorFlow Modeling Pipeline Using TensorFlow Datasets and TensorBoard

⦿ Overview : How can data scientists add rigor and structure to their modeling experiments?⦿ Our contribution : This blog post discusses approaches and tools that can help machine learning modelers organize, evaluate and record their experiments with algorithms and configurations. The post explores tools though a practical demonstration in malaria image-detection use case.⦿ Resources:
  - Blog

• Writings
• Team
• Home

25 NTC
April 2025
Baltimore, MD

AI For Good
February 2025
Stanford University, CA