# Jacob Davies

Incoming FDSE @ Palantir | AI Researcher

[Contact Me](mailto:jacobrhys04@gmail.com) · [LinkedIn](https://www.linkedin.com/in/jacob-davies-654399271/)

## Projects

### AI Weather Prediction (AIWP)

*Undergraduate Dissertation - University of Birmingham*

Improved ML-based weather prediction for stochastic sub-seasonal forecasts. Trained ensemble models over ERA5 reanalysis; evaluated skill vs. IFS benchmarks at 2–6 week horizons.

[Blog](https://jacob-davies.com/blog/ai-weather-prediction/) · [Paper](https://jacob-davies.com/papers/ai-weather-prediction/)

### Satellite-Based AI Vessel Detection

*Postgraduate Dissertation - University of Warwick*

Edge-deployed vessel detection over Planet Labs SuperDove imagery. Detects and tracks maritime targets under cloud cover, latency-bounded for on-satellite inference.

[Blog](https://jacob-davies.com/blog/satellite-vessel-detection/) · [Paper](https://jacob-davies.com/papers/satellite-vessel-detection/)

### Climate Birthday

*Personal Project*

Which day of the year now has the weather your birthday used to have? Per-city daily climatologies from ERA5 reanalysis, pre-industrial vs today, inverted to find your climate birthday - and the days whose weather no longer exists.

[Live demo](https://jacob-davies.com/climate-birthday/) · [Blog](https://jacob-davies.com/blog/climate-birthday/)

### Structure-Taught Protein Models

*Research*

Can a small protein model learn structure without ever predicting amino acids? I pretrained eleven 7.4M-parameter encoders on the same 486M-token budget, comparing masked language modelling with latent prediction against a self-distilled target and against frozen ESM-IF1 structure embeddings. The self-distilled version collapsed as predicted and latent prediction alone did not beat masked language modelling, but distilling the structure embeddings directly did on every task. The demo runs two of the encoders in your browser and predicts secondary structure for any sequence.

[Live demo](https://jacob-davies.com/protein/) · [Blog](https://jacob-davies.com/blog/structure-taught-protein-models/) · [Paper](https://jacob-davies.com/papers/structure-taught-protein-models/)

## CV

### Employment

#### Palantir — Forward Deployed Software Engineer

*Starting November 2026*

Starting in November 2026, I will be working at Palantir as a Forward Deployed Software Engineer.

#### Quantcast — Machine Learning Engineer Intern

*Jul – Sept 2024*

During summer 2024, I interned at Quantcast in their Machine Learning Team. Rather than giving me an isolated project, they got me to work on adding functionality to their household clustering algorithm that deduces which devices are contained within a household. I completed the 12 week project in 7 weeks; for the last 5 I worked on improving the algorithm and saw a 1000x increase in stability.

#### WRJ Solutions — Co-Founder

*Jun 2022 – Dec 2023*

Co-founded a startup with two other students. Developed a system that takes in call centre transcripts, analyses the theme and individual topics of each call, and summarises the results in a readable format for internal use. Sold the system to Ericsson, the largest producer of mobile telecommunications equipment.

#### Triumph Motorcycles — Work Experience - Triumph Design Award

*Jul 2023*

Work experience awarded for winning the Triumph Design Award. I was based in the Power Unit Engineering department, but got to experience every department in the company and sit in on meetings. A great insight into how a large company works, and how changes and decisions are made.

### Education

#### University of Warwick — Postgraduate - AI

*2025 – 2026*

Fundamentals of AI research, Programming for AI, Fundamentals of AI and data mining, AI and deep learning, Applied statistics for AI, Ethical AI implementation, AI for industry.

#### University of Birmingham — BSc Computer Science and Artificial Intelligence

*2022 – 2025*

Neural computation, Data analysis, Natural Language processing, Dependable and distributed systems, Game theory. Dissertation: Improving Machine learning-based weather prediction techniques for suitability in stochastic sub-seasonal forecasts - 2.1

#### Ysgol Uwchradd Aberteifi — A-levels

*2020 – 2022*

Mathematics, Physics, Product Design - A*A*A*