Applied AI engineer / Philadelphia, PA

EmmettHintz.

I build AI tools
people can rely on.

I build document-processing systems, voice tools, and workflow software at Foley, from the first requirements to production.

Explore my work

Currently at Foley

01 / Selected work

A few examples of
what I've built.

At Foley, I build software for document processing, customer analytics, and background screening. Here are three of those projects.

01Driver document processing

Order fromthe paperwork.

Splitting driver packets and matching documents to the right driver, with uncertain results sent for review.

OCR · Bedrock vision · Kubernetes

50–75

pages per driver packet.

Automating driver document intake.

Operations teams were splitting 50- to 75-page driver packets by hand, then matching each document to the right driver. I led the platform from stakeholder requirements through deployment and maintenance. It uses layout rules for straightforward splits, a vision model for the harder cases, and human review when the result is uncertain.

Cut months from onboarding in the worst cases for customers with large driver populations.

50 to 75-page driver packets / OCR · Bedrock vision · Kubernetes / Confidence checks · human review

02Customer health

Readingthe signals.

A customer-health scorecard that helps Customer Success decide which accounts to review first.

Python · SQL · point-in-time panels

months of account history.

Prioritizing customer account reviews.

Customer Success needed a way to prioritize account reviews. I built an explainable scorecard from 43 months of history, testing each score against the information that would have been available at the time. Two signals people expected to matter did not hold up in testing, so I ruled them out before rollout.

In one tested segment, the highest-risk 20% of accounts captured about 55% of later churn events.

43 months of account history / Python · SQL · point-in-time panels / Explainable scores · human judgment

03Court-record review

Followthe evidence.

Searching federal courts and tracing findings to their source pages, with final decisions left to the reviewer.

FastAPI · Next.js · PostgreSQL

workspace for court searches.

Federal court records, ready for review.

Screening teams had to search court by court, sort out people with similar names, review PDFs, and re-enter what they found. I built a workspace that searches dozens of federal district courts and links extracted facts to their source pages. Rules determine which records need review, and a person makes the final compliance decision. The search and review workflow also works without the model.

Built a pilot-ready platform with page citations for every extracted fact and a human reviewer making the final decision.

Federal court records / FastAPI · Next.js · PostgreSQL / Page citations · spend limits · human final authority

Public projects / GitHub

More
projects.

Machine learning and research projects, with the code on GitHub.

01 / Reinforcement learning

AlphaZero Chess Bot

A PyTorch chess bot trained through self-play, Monte Carlo tree search, and neural network evaluation.

View repository

02 / ML research

miRNA Progression Models

Machine-learning research on miRNA expression data that identified four biomarkers associated with prostate-cancer progression.

View repository

03 / Computer vision

Chess Position Recognition

Reads a hand-drawn chessboard and converts the position into FEN notation.

View repository

04 / AI product

Job Finder

Searches multiple job boards, extracts structured listing data, and removes duplicates.

View repository

02 / Experience

Where I've worked and what I've done.

Jul 2025 to present

Foley

Applied AI Engineer

I build and maintain document-processing, voice, and workflow systems. Recent work includes automating driver onboarding, restoring missed customer callbacks, and finding gaps in legacy billing logic.

Python · TypeScript · PostgreSQL · LLMs · OCR

Remote

Aug 2024 to May 2025

Colgate University

Networking Research Assistant

I built a research pipeline that translated raw IP data into readable domain patterns for network-topology analysis.

Python · pattern recognition · data pipelines

Hamilton, NY

May to Aug 2024

Colgate University

Machine Learning Research Assistant

I built machine-learning workflows for gene-expression research and identified four biomarkers associated with prostate-cancer progression.

PyTorch · XGBoost · scikit-learn · research

Hamilton, NY

Oct to Nov 2023

Tewke

Software Engineer Intern

I designed and shipped a responsive dashboard for customer support reporting, monitoring, and data visualization.

React · TypeScript · product engineering

London, UK

03 / How I work

I build tools that save people time and let them check the work.

I work mainly in Python and TypeScript, with SQL and PostgreSQL for data systems. My recent projects have included OCR, structured extraction, LLM integrations, retrieval, classification, and workflow automation. I build in validation, keep a record of how results were produced, and route uncertain cases to a reviewer.

01Document processing and OCR

02LLM integrations and retrieval

03Workflow automation

04Backend and data systems

05Review tools and handoffs

06Tracing, evaluation, and safeguards

EmmettHintz.

A few examples ofwhat I've built.

Order fromthe paperwork.

Readingthe signals.

Followthe evidence.

Moreprojects.

AlphaZero Chess Bot

miRNA Progression Models

Chess Position Recognition

Job Finder

Where I've worked and what I've done.

Applied AI Engineer

Networking Research Assistant

Machine Learning Research Assistant

Software Engineer Intern

A few examples of
what I've built.

More
projects.