About Jeff

Hi. I'm Jeff. I am interested in how humans use computers, and the interfaces between them.

I have been working in this area for about 30 years, prior to the general availability of the World Wide Web. Since those early days, I have worked as an interface designer, graphic artist, illustrator, web developer, software engineer, and user experience designer.

Much of this work has revolved around the Web, though this is not by any means my sole interest. For the last several years I have been working in the machine learning / AI space.

One question I have been asking since I become involved with AI is, “What is the user interface for AI?”. With the advent of Large Language Models, this question has become even more critical.

In a world of Generative AI, how will humans interact with systems that become more and more inscrutable? How do we design systems that produce good outcomes when the underlying system is largely non-deterministic? These are the questions on my mind lately.

Portfolio

Recent work, in reverse chronology. Much of this work was produced by teams of people, in which I played a role. These roles include: engineer, designer, UX designer, leader, manager, or any combination of these. I will attempt to clarify my role in each case.

Lawful Good

In 2025, I was the founder and principal designer at Infinite Tape, a startup that builds AI-powered tools for professionals and creatives.

One of the projects I developed was called Lawful Good. This project is ongoing as of this writing. Lawful Good is an AI Agent designed to help independent lawyers manage documents more effectively using a RAG-based approach.

Date
2025
Company
Infinite Tape
Roles
Engineer, Designer
Stack
React, TypeScript, Python, FastAPI, LangChain, OpenAI, Gemini, PostgreSQL, ChromaDB, Google Cloud Platform
Lawful Good Workflow
Lawful Good Empty State

H2O Actions

H2O Actions was the brainchild of Jiri Puc, an AI Engineer at H2O.ai. The idea behind this project was to create an AI Agent that could be used to automate tasks in the H2O AI Cloud, using a natural language interface. Users would be able to create experiments, perform data exploration, launch AI Engines for Driverless AI, and more, all by interacting with the AI Agent.

Date
2024
Company
H2O.ai
Roles
Designer
Stack
React, TypeScript, Python, LangChain, h2oGPT, OpenAI, Kubernetes
H2O Actions Action Selection Thumbnail
H2O Actions Navigation Thumbnail
H2O Actions Empty State

AI Governance

AI Governance is an important part of managing the deployment of AI models in the enterprise. Understanding model risk and health is critical for business leaders to understand their risk exposure.

The AI Governance project at H2O.ai was designed to help Risk Management Officers understand at the highest level what models they have deployed, and to quantify the risks associated with these models.

Date
2024
Company
H2O.ai
Roles
Designer, Engineer
Stack
React, TypeScript, Java, PostgreSQL
AI Governance Model Health Thumbnail
AI Governance Models

Document AI Dashboard

H2O.ai's Document AI is designed to help data scientists create data extraction pipelines from large quantities of documents, using combinations of models ranging from OCR, layout models, and LLMs.

Date
2023
Company
H2O.ai
Roles
Designer, Engineer
Stack
React, TypeScript, Java, PostgreSQL
H2O Document AI Pipelines Thumbnail
H2O Document AI Pipelines Expanded Thumbnail
H2O Document AI Dashboard

H2O AI Cloud

In 2019, H2O.ai embarked on an ambitious project to build a cloud-based platform for AI. Named simply H2O AI Cloud, the project was intended to bring together the best of H2O's AI products into a single, unified platform. Key components included H2O's flagship products, Driverless AI and H2O-3, and a suite of cloud-based products, such as Hydrogen Torch, MLOps, Feature Store, and Document AI. The idea was to allow H2O's customers to build apps using these tools and H2O's low-code app framework, Wave, and deploy them on their own App Store instance.

I joined the project in 2020 after it had begun development. I was initially tasked with leading the UI engineering effort. Eventually, I also became involved with leading the UX team.

Date
2020-2024
Company
H2O.ai
Roles
Lead Engineer, UX Manager
Stack
React, TypeScript, Kubernetes, Go, Protocol Buffers, Python, Java, PostgreSQL
App Store Detail Thumbnail
AI Engine Manager Thumbnail
AI Engine Manager Selection Thumbnail
AI Engine Manager Configuration Thumbnail
AI Engine Manager Logging Services Thumbnail
H2O Cloud App Store Thumbnail

PwC

In 2016, PwC (PricewaterhouseCoopers) engaged with H2O.ai to build AI-powered auditing applications. Initially, my role was to design and build these applications in collaboration with H2O.ai's data science team. Since the PwC group we were working with was based in London, the leadership team there eventually chose to work with a design team based in London. Thereafter, my role was primarily as Lead UI Engineer, but I occasionally gave UX feedback to the London-based design team.

Between 2016 and 2021, I worked on a number of projects including Journals.ai, GL.ai, Fit.ai, Controller.ai, Cash.ai, and Audit.ai. PwC is quite private with their internal projects, so much of this work I am not able to share publicly. What I am able to share, I have included below.

Cash.ai

PwC's Cash.ai application is used internally by PwC audit teams to accelerate the audit process. It uses machine learning, machine vision, and other AI processes to quickly and accurately scan documents, extract their data, and correlate this data with the audited journals.

Cash.ai is one module of a larger suite of applications called Audit.ai. Each module focuses on one aspect of the audit process. Other modules include Account Payable, Accounts Receivable, Property, Plant, & Equipment (PP&E), etc.

Date
2017
Company
PwC
Roles
Engineer, UX Guidance
Stack
Angular, TypeScript, Java, Spring Boot, PostgreSQL
Cash.ai Document Thumbnail
Cash.ai Video Thumbnail
PwC's Cash.ai

GL.ai

PwC's GL.ai application is an AI tool which uses anomaly detection to analyze entire ledgers using machine learning, providing a great deal more efficacy than the traditional statistical sampling method, which only observes a small portion of the overall general ledger.

While the AI handles the analysis, the centerpiece of the user experience are data visualizations which display the results of the models. These visualizations make it easy for auditors to spot inconsistencies and unexpected patterns.

Date
2019
Company
PwC
Roles
Engineer, Data Visualization, UX Guidance
Stack
Angular, TypeScript, D3.js, Java, Spring Boot, PostgreSQL
GL.ai Video Thumbnail
PwC's Cash.ai