Patrick Frenett

Experience

London, UK

AI EngineerJuly 2024 – Sept 2025

Led the development of agentic search, building extraction, indexing, retrieval, and generation components. Wrote the entire evaluation suite.
Various smaller projects: Methods for PowerPoint and Excel document understanding, editing, and generation. Improved PDF citation and bounding box extraction. Built out deep research modules. Prototyped generic entity extraction and linking modules with arbitrary ontologies.
Worked with finance, legal, insurance, and logistics customers to build out custom solutions. Led sales calls and demos for customers. Conducted product research and development.

London, UK

Founder and Technical LeadOct 2023 – Present

Built a mystery guest booking and review platform for restaurants, bars, and gyms. Work with over 50 sites across 8 countries. $50k ARR.
Co-founded with a friend in hospitality, takes up around 3 hours of my time per week.

London, UK

Senior Data ScientistSept 2022 – July 2024

Led the internal alpha development program for prototyping new LLM-based features. Rapidly developed internal apps shared with internal and external stakeholders.
Led the development of a fine-tuned, on-premise generative longform question-answering model with novel citation and source tracking. Included research, development of synthetic data pipelines, model fine-tuning, RLHF, benchmarking, and deployment. The trained 7B model matched GPT-4 (SOTA at the time) in terms of factual consistency and hallucination reduction.
Deployed production models and data pipelines for entity extraction, entity linking, event extraction, and document summarisation.

London, UK

Data ScientistOct 2021 – Sept 2022

Built language models used to identify skill gaps between job postings and UK apprenticeship standards. Developed client-facing applications to communicate results.
Developed a product recommendation model using pretrained embeddings to determine similar products for e-commerce merchants facing cold starts.
Trained and deployed a computer vision model for COVID-19 test cassette recognition using AutoML tools.
Led the team writing quotes for 6-figure data science contracts.

Estonia, Germany, Kenya, Uganda

Thomas J. Watson FellowAug 2019 – Aug 2020

One of 41 fellows selected for a year of funded independent study. Project: “Innovation, Automation, Motivation”.
Visited 11 countries across Africa and Europe studying societal impacts of automation, focusing on the psychology of motivation and what makes work meaningful.

San Francisco, USA

Software Engineering InternApril 2017 – Sept 2018

Built a HIPAA compliant database to track patient data, incorporating 5+ years of survey data. Reduced production cycle by ~5 days per quarter.
Developed a client-facing platform for custom research, including a conjoint simulator offered as a major component of projects selling for up to $100,000 each.
Built a web application for analysing email campaigns.

Feb 2026 – Present

Open source tool for versioning and collaborating on non-code data sets (pdf, pptx, etc). Built for swarms of agents contributing to the same document sets. Git-like provenance of memories derived from the document sets.
Built as a CLI tool and associated LLM skill.

Jan 2026

Podcast ingestion + analysis tool. Available as a CLI tool to give command line agents access to podcast data.
Transcription, diarisation, topic modelling, hybrid search, recommendation system, chat system.

Sept 2025 – Dec 2025

AI workspace for VC analysts. IDE-shaped UI for non technical users interacting with non-code files and folders.
Built out extraction, indexing, retrieval, and generation components.
Conducted user research and onboarding of beta users. Decided to pivot away from the project and towards tools for existing similar solutions (Claude Cowork, OpenClaw, Zo, etc).

London, UK

M.Sc Computing (Artificial Intelligence and Machine Learning)Sept 2020 – Sept 2021

Distinction (74% Examination Average, 78% Dissertation)
Dissertation: “Neuro-Symbolic Learning with Logically Constrained Conditional Autoencoders”
Independent Study: “Semidefinite Programming Approaches to Neural Network Verification”

Amherst, MA, USA

B.A. Computer Science, Mathematics, Statistics (Triple Major)Sept 2015 – May 2019