Science · AI · Data Science

Thinking out loud
about science and AI.

Some of the biggest changes happening in science right now are being driven by AI and data science - and I think they should be accessible to everyone, not just people with a machine learning background. I write here about what's actually going on: how AI is transforming industries, the most significant implications for drug discovery and beyond, what models exist and what they've already unlocked - and how to actually use them. I also write about the science of working in science - teams, collaboration, and what it takes to do this well in practice. A mix of opinion, industry analysis, technical posts, and hands-on tutorials - written from someone who uses these tools in real research, not just writes about them.

Opinion Published · March 2026 Great science needs more than scientists Diversity of background does not just make teams more pleasant to work in. It makes the science better. I have seen the same thing play out across very different settings - and I think the reason is worth examining carefully. Read post → Industry & Research Published · March 2026 How AI is changing drug discovery - and what the evidence actually says The industry is reorganising fast. AI-designed drugs are reaching clinical trials. Pharma companies are embedding AI infrastructure directly into their R&D pipelines. This post covers what is actually happening, what the evidence shows, and the honest open questions about whether this means better medicines. Read post →
Opinion Coming soon What makes a team actually work - on feedback, trust, and creating space for honest challenge The leadership questions that rarely get asked directly: how to build teams where people feel safe enough to disagree, how to give feedback that lands well, and why psychological safety is not a soft metric.
Technical Coming soon What NVIDIA's BioNeMo actually means for wet-lab scientists A practical look at what BioNeMo, DiffDock, and the broader NVIDIA biology stack make possible - and what still needs a specialist to get right.
Technical Coming soon Using ESM-2 to predict protein fitness - a walkthrough with real data A hands-on look at Meta's protein language model: what it actually does, how to run it, and how to interpret the output in a biological context.
Technical Coming soon Unsupervised ML for immune cell population discovery in RNA-seq data From alignment to UMAP - how to use clustering and dimensionality reduction to define cell populations without prior labels, and what to watch out for when you do.
Tutorials & Guides Coming soon RNA-seq from scratch - a practical guide for biologists who have never touched the command line A step-by-step walkthrough of a full RNA-seq pipeline - from raw FASTQ files to differential expression results - written for researchers with domain knowledge but no computational background.
Tutorials & Guides Coming soon What AlphaFold actually gives you - and how to make sense of it A plain-language guide to interpreting AlphaFold output: confidence scores, what pLDDT means in practice, and how to decide whether a predicted structure is useful for your question.