Phil Palmer

Phil Palmer

Co-founder

safely.bio

Hi there!

I'm co-founder of safely.bio, where we're building a verification system for the biotech industry.

Previously, I designed a broad-spectrum coronavirus vaccine during my PhD at Cambridge, built metagenomics pipelines for threat detection at MIT, and was employee ~10 at Lifebit Biotech through a $7.5 million Series A. I've also worked with biosecurity think tanks such as the Nuclear Threat Initiative and the Centre for Long-Term Resilience on AI-enabled biological risks, and helped scale the Cambridge Biosecurity Hub from 0 to 30+ members.

Thinking about biosecurity, AI × bio, and startups.

Publications

Everything else. Full listing on Google Scholar

Safety First: Input Screening for Protein Design Tools

bioRxiv preprint, 2026 · Nuclear Threat Initiative

P. Palmer, N. Teran, N. Wheeler, J.M. Yassif

AI protein design tools can produce novel sequences that similarity-based screening misses. BioScreen screens the target instead: a database of 14,541 potentially harmful human proteoforms, classified by risk level, combined with embedding-based screening using the ESM-C protein language model. It reaches F1 > 97% on variants of known targets and tells benign mutations apart from damaging ones where BLASTP does not.

Preprint PDF Code
Input screening workflow for protein design tools

A computational method to design broad-spectrum T cell-inducing vaccines applied to Betacoronaviruses

Cell Reports Methods, 2026

P. Palmer, S. Fedosyuk, S. Parthasarathy, J. Holbrook, C. George, L. O'Reilly, L. Wiegand, G.W. Carnell, J.L. Heeney, S. Vishwanath

Spectravax designs vaccines that account for genetic diversity in both the human host and the pathogen. A single designed nucleocapsid antigen raised immune responses in mice to SARS-CoV, SARS-CoV-2 and MERS-CoV — the three coronaviruses behind every major outbreak since 2002.

Paper PDF Preprint Code
Spectravax graphical abstract

Global Risk Index for AI-enabled Biological Tools

The Centre for Long-Term Resilience & RAND Europe, 2025

A global assessment of the dual-use risks posed by narrow AI-enabled biological tools, built to inform policy on reducing the risk of AI-enabled biothreats. I contributed tool evaluations as a consultant.

Report
Global Risk Index for AI-enabled Biological Tools report cover

Computationally designed haemagglutinin with nanocage plug-and-display elicits pan-H5 influenza vaccine responses

Emerging Microbes & Infections, 2025

C.Q. Huang, R.A. Hills, G.W. Carnell, S. Vishwanath, E.T. Aguinam, A.C.Y. Chan, P. Palmer, L. O'Reilly, P. Tonks, N. Temperton, S.D.W. Frost, L.S. Tiley, M.R. Howarth, J.L. Heeney

A computationally designed H5 antigen displayed multivalently on the mi3 nanocage. At low doses it elicits cross-clade neutralising antibody and T cell responses against diverse H5 bird flu strains — a scalable candidate for pre-pandemic stockpiling.

Paper PDF Preprint
Haemagglutinin antigen displayed on a nanocage

MYC regulates a pan-cancer network of co-expressed oncogenic splicing factors

Cell Reports, 2022

L. Urbanski, M. Brugiolo, S. Park, B.L. Angarola, N.K. Leclair, M. Yurieva, P. Palmer, S.K. Sahu, O. Anczuków

A network of splicing factors co-regulated with MYC across 33 tumour types, yielding a splicing signature that correlates with survival. While at Lifebit I built the Nextflow pipeline behind the large-scale splicing analysis.

Paper PDF Code
MYC splicing factor network graphical abstract
🏆 Winner · def/acc Hackathon, London

BioSecure: Know-your-customer system for DNA synthesis companies

BlueDot Impact & Apart Research, 2025

P. Palmer, R. Teo, M. Karlubík, J. Harris

A modular customer-screening framework for DNA synthesis companies, built in 48 hours. It combines ID verification, sanctions screening, institutional and publication checks, and protein language model sequence screening into one cumulative risk assessment that supports human decisions rather than replacing them.

Demo PDF Project Hackathon
BioSecure demo application
🏆 Winner · AI × Bio Hackathon, London

SplitScreen: Benchmarking obfuscated split-order detection methods

BlueDot Impact, Apart Research & Cambridge Biosecurity Hub, 2026

A. Jones, P. Palmer, P. Garg

DNA synthesis screening works on single oligos, so an adversary can split a hazardous gene across many short fragments that each pass. We built an open-source benchmark for split-order detection: a data generator with configurable obfuscation, six detection methods, and an evaluation harness. Alignment-based methods stayed robust under codon substitution where exact-match methods broke down.

PDF Code Hackathon
Presenting SplitScreen at the AI x Bio hackathon

AutoPlate: Rapid dose-response curve analysis for biological assays

Frontiers in Immunology, 2022

P. Palmer, J.M.M. Del Rosario, K.A.S. da Costa, G.W. Carnell, C.Q. Huang, J.L. Heeney, N.J. Temperton, D.A. Wells

An open-source R Shiny app that automates dose-response curve fitting for immunoassays such as the pseudotype microneutralisation assay, cutting analysis time during COVID-19 and influenza work.

Paper PDF Code
AutoPlate assay workflow

Neutralisation hierarchy of SARS-CoV-2 variants of concern using standardised, quantitative neutralisation assays reveals a correlation with disease severity

Frontiers in Immunology, 2022

D. Cantoni, M. Mayora-Neto, A. Nadesalingam, D.A. Wells, G.W. Carnell, L. Ohlendorf, M. Ferrari, P. Palmer, et al.

Benchmarks how well first-wave convalescent sera neutralise each SARS-CoV-2 variant of concern, reported in international units, and finds that more severe COVID-19 corresponded to broader neutralisation.

Paper PDF
Standardising neutralisation assays across laboratories

Analysis of serological biomarkers of SARS-CoV-2 infection in convalescent samples from severe, moderate and mild COVID-19 cases

Frontiers in Immunology, 2021

J. Castillo-Olivares, D.A. Wells, M. Ferrari, A.C.Y. Chan, P. Smith, A. Nadesalingam, M. Paloniemi, G.W. Carnell, L. Ohlendorf, D. Cantoni, M. Mayora-Neto, P. Palmer, et al.

Compares neutralising and binding antibody assays against the WHO international standard to find biomarkers that distinguish severe from mild infection, as candidate correlates of protection.

Paper PDF
Correlation matrix of serological biomarkers

Experience

Co-Founder · safely.bio

Apr 2026 – Present · Cambridge, UK

  • Building a verification system for the biotech industry.

Research Lead · Cambridge Biosecurity Hub

2021 – Present · Cambridge, UK

  • Scaled the hub from 0 to 30+ members, helped run the weekly reading group, and mentored two graduate students on a biosecurity project quantifying pandemic risks.

Technical Consultant · Nuclear Threat Initiative

Jul 2025 – Mar 2026

  • Developed BioScreen, a biosecurity screening method for AI protein design tools.

Founder in Residence · Ambitious Impact

Jan 2026 – Mar 2026 · London, UK

  • Selected as one of 20 participants from 1,000+ applicants for the Founding to Give programme.
  • Explored and validated high-impact venture opportunities across biotech.

5050 Fellow, UK Cohort 2 · Fifty Years

Oct 2025 – Dec 2025

  • Building a biosecurity startup.

Consultant · The Centre for Long-Term Resilience

Feb 2025 – Mar 2025

  • Evaluated narrow AI-enabled biological tools to assess their dual-use risks as part of the CLTR–RAND Risk Index project, informing policy efforts to reduce the risk of AI-enabled biothreats.

PhD in Biology · University of Cambridge

Oct 2020 – Feb 2025 · Cambridge, UK

  • Lab of Viral Zoonotics, supervised by Prof. Jonathan Heeney.
  • Developed computational methods to design broad-spectrum vaccines and antibodies against antigenically diverse pathogens.

Visiting Researcher · MIT Media Lab

Oct 2022 – Jan 2023 · Boston, MA

  • Evaluated and improved a general-purpose metagenomics pipeline for the Nucleic Acid Observatory, which aims to build a reliable early detection system for catastrophic biothreats.
  • Supervised by Prof. Kevin Esvelt and Dr Mike McLaren.

Bioinformatician · Lifebit

Jul 2018 – Jul 2020 · London, UK

  • Joined as employee ~10 shortly after the seed round; contributed to growth to ~30 employees and a $7.5M Series A, including delivery of the Genomics England Research Platform for COVID-19 response.
  • Built bioinformatics pipelines adopted by leading genetics firms including Nebula Genomics and Sano Genetics, analysing data from thousands of customers.

BSc Biology, First Class Honours · University of Southampton

2015 – 2018 · Southampton, UK

Contact

LinkedIn