Sam McKaylin

Jul 2026 – Present

I set technical strategy and built the GCP and AWS infrastructure for a paediatric imaging AI company. I lead a foundation model trained on approximately 1.2 million paediatric radiographs across four cohorts and designed a two-stage diagnostic system combining YOLO detection with a fine-tuned MedGemma model for automated clinical reports. I built federated learning infrastructure for NHS deployment, manage NHS relationships and data access, contribute to regulatory strategy, and project-manage delivery.

Jan 2025 – Present

I lead scientific strategy for a precision oncology clinical decision-support software company, building products and teams, derisking science and technology, and navigating regulatory strategy and procurement. I built AI-enabled electronic molecular tumour board software, fine-tuned LLMs to extract evidence from more than 170,000 papers and trial records, and developed pathway-based response classifiers validated by expert oncologist panels. I established collaborations with MedStar Health and Yale, integrated NHS genomic laboratory data, and project-manage engineering delivery.

Oct 2024 – Present

I perform investment due diligence on and maintain relationships with dozens of top-tier biotech and healthcare venture funds across the US, Canada, and UK for allocators overseeing more than $750B. I assess portfolio companies’ science and technology alongside team, strategy, track record, and quantitative factors; support LP clients on co-investments and fund commitments; and author white papers on sector trends and opportunities.

May 2023 – Sep 2024

I developed pathway-constrained Transformers from scratch using single-cell CRISPR screens, more than 1.5 million dose-response curves, transcriptomics, and whole-genome sequencing. The model outperformed DeepSynergy on every held-out split and demonstrated that its reported performance was inflated by data leakage. My findings derisked oncology programs and supported an in-licensing partnership with Astex Pharmaceuticals and acquisition of an asset now in Phase II. I also replaced exhaustive dose-response matrices with a checkerboard design, roughly halving wells per drug combination, mentored two junior ML team members, and trained a 20-person wet-lab team in Python.

Sep 2021 – Apr 2023

I built machine learning models, including Transformers, CNNs, and ChemBERTa, to identify small molecules that modulate biomolecular condensates. I also developed high-throughput image-analysis pipelines for screening.

MPhil in Zoology, Developmental Biology

2020 – 2021

Distinction

Research project in the Eco-Evo-Devo lab exploring craniofacial evolution using spatial transcriptomics and morphometrics.

BSc in Biological Sciences, Biological Sciences

2016 – 2020

First Class Honours, Dean’s List

First class degree (4.0 GPA), thesis explored the spread of ideas through literature using natural language processing and machine learning.