docs: propose agentic discovery + inference engineering additions to CV

.gitignore

@@ -9,3 +9,6 @@ _site/
 # Jekyll metadata
 .jekyll-metadata
 
+
+# Vale linter — fetched via `vale sync`, not for commit
+.vale-styles/

.markdownlint.jsonc

@@ -0,0 +1,11 @@
+{
+  // CV and prose use LaTeX-style ventilated prose: one sentence per line.
+  // Disable hard line-length cap; sentences run as long as they need to.
+  "MD013": false,
+  // Ordered list flexibility for publication numbering.
+  "MD029": false,
+  // Allow inline HTML (we use angle brackets for emails).
+  "MD033": false,
+  // Don't require H1 at top of file (Jekyll layout supplies title).
+  "MD041": false
+}

.vale.ini

@@ -0,0 +1,7 @@
+StylesPath = .vale-styles
+MinAlertLevel = warning
+Packages = write-good, proselint
+Vocab = ctr26
+
+[*.md]
+BasedOnStyles = write-good, proselint

docs/sanger-gdm-cover-letter.md

@@ -0,0 +1,81 @@
+# Cover Letter: Google DeepMind Fellow in Genomics and AI
+
+Dr Craig T. Russell
+London, UK · <craig.russell.phd@gmail.com> · [ctr26.github.io](https://ctr26.github.io)
+
+To the Fellowship Review Panel,
+Wellcome Sanger Institute & Google DeepMind Academic Fellowship Programme
+
+---
+
+## Why Sanger, and why now
+
+This Fellowship sits where my work is heading.
+I want to take the multi-modal foundation-model approaches I have built in industry and ground them in the world's richest genomics environment, to deliver fundamental biological insight.
+Sanger is the only place in the UK where atlas-scale single-cell data (Cellular Genomics), a programme committed to predictive biology (Generative Genomics), and a strategic embrace of AI for science share one building.
+I want to spend three years here.
+
+I am Senior Machine Learning Scientist at Valence Labs / Recursion Pharmaceuticals.
+I lead components of the Virtual Cell initiative, fine-tuning multi-modal LLMs over knowledge graphs, free text, transcriptomic and phenotypic imaging data to predict cell state and drug response.
+I co-developed TxPert (arXiv:2505.14919), a state-of-the-art transcriptomic perturbation predictor, and contributed to the Boltz2 proteome-scale virtual screening pipeline.
+Before Recursion I led AI engineering at EMBL-EBI's BioImage Archive and Uhlmann Group, supervised six PhD students, and shipped `bioimage_embed` (self-supervised learning for biological images) and `shape_embed` (cell-shape representation, arXiv:2507.01009) as community resources.
+My PhD (Cambridge, 2018) was an instrument-building project: I designed and built a light-sheet microscope from scratch.
+I am as comfortable at the data-generation end of the stack as at the modelling end.
+
+## My Grand Challenges
+
+I would primarily target Grand Challenge 4: *Harnessing multimodal generative AI to study the interplay of genetics with spatial transcriptomics, proteomics and morphology in human tissues*, led by Mo Lotfollahi and Muzlifah Haniffa.
+The downstream goal of predicting genetic variation from routine clinical pathology slides sits exactly at the intersection of the multi-modal foundation models I have built at Recursion (TxPert; the Virtual Cell initiative integrating knowledge graphs, transcriptomics, free text and phenomics; Boltz2 components) and the imaging methods I shipped at EMBL-EBI (`bioimage_embed`, `shape_embed`, MIFA).
+Sanger's >200 million curated cells across H&E, Xenium, scRNA-seq and ATAC are an unparalleled training corpus for cross-modal generative architectures.
+
+I would secondarily consider Grand Challenge 5 (*A Foundation sequence model of the Cis-Regulatory Code in developing human tissues*, Lotfollahi/Taipale) and Grand Challenge 2 (*Predictive and Mechanistic AI for Longitudinal Multi-Omics in Complex Disease*, Anderson) as natural adjacencies, but Challenge 4 is where my unique skills land best.
+
+The technical novelty I would bring has four parts.
+First, integrate Sanger's H&E + Xenium + scRNA-seq + ATAC corpora under a single multi-expert architecture with shared latent space, with knowledge graphs as structural priors and curated literature as semantic anchors.
+The architectural lineage is the multi-modal LLM-over-omics work I have shipped at Valence; the data scale is what makes Sanger the right home for it.
+Second, evaluate not only held-out reconstruction but counterfactual prediction.
+Does the model give scientifically reasonable answers when asked "what would the spatial transcriptome and morphology look like for this individual's variant in this tissue?".
+Third, explicitly evaluate and correct for ancestry-shift and tissue-shift in the latent space, so that variant-to-pathology predictions are anchored in data the model has actually seen rather than extrapolated unsafely from a single reference cohort.
+This connects directly to my outreach plan below.
+
+Fourth is the part I would most like to push the field on: build an agentic discovery layer on top of the foundation model, walled in by inference-time engineering so that hallucination is contained to scientifically tolerable failure modes.
+Naive agentic LLMs in biology fail loudly.
+The agent invents a gene name, fabricates a citation, proposes a perturbation that exists in no cell line.
+The engineering response is to refuse free generation over the things that matter.
+Constrained decoding restricted to a closed vocabulary of valid Ensembl IDs, HGNC symbols, and ChEMBL identifiers; tool-call verification against Sanger's own structured resources (Cellular Genomics atlases, Tree of Life genomes, the human variant catalogues), so that every claim round-trips through real data before the agent emits it; retrieval grounding to a curated literature corpus with citation-level attribution; verifier-model ensembles that score each hypothesis for biological plausibility before it leaves the agent.
+The result is an agent that proposes follow-up experiments (variants to validate, tissues to oversample, perturbations to run) but only inside a sandbox where every leaf claim is checkable.
+This is the missing piece between an LLM that talks about biology and an autonomous research collaborator.
+GDM mentorship is most directly useful here: the inference-engineering and constrained-generation work at DeepMind on AlphaFold's confidence calibration and the AlphaProof verifier loop is the lineage I want to bring into genomics.
+
+Three deliverables in year one.
+(i) A Sanger-anchored multimodal foundation model trained jointly on Cellular Genomics atlases (H&E + Xenium + scRNA-seq + ATAC), with public benchmarks for cross-modal counterfactual prediction.
+(ii) A variant-to-pathology benchmark co-designed with Lotfollahi and Haniffa's groups that scores models on ancestry-stratified validation cohorts rather than convenience samples.
+(iii) An open-source agentic discovery harness with built-in hallucination suppression (constrained decoding, tool-call verification, retrieval grounding, verifier ensembles) that other Sanger researchers can wrap around their own foundation models.
+Year two onward extends this to the cis-regulatory code (Challenge 5) and to longitudinal cohorts (Challenge 2) where the same foundation-model backbone applies.
+
+For secondary supervisor I suggest Prof. Pietro Liò (Cambridge, Computer Laboratory) for his work on graph foundation models for biology, with Dr Virginie Uhlmann (EMBL Barcelona, formerly EMBL-EBI) as a complementary option from my prior collaboration on bioimage representation learning.
+Neither is a precondition; I am open to whichever AI/ML academic the panel matches me to.
+
+## Outreach: capacity-building for Genomics and AI in LMICs
+
+The outreach fund is what I most want to use this Fellowship for.
+I will run a recurring "Foundation Models for Genomics" workshop series, twice a year at LMIC partner institutions, one week each, producing graduates who can fine-tune and evaluate ML models on local genomic data.
+The teaching pattern is the one I already run at EMBL-EBI for 40+ researchers a year, adapted for compute-constrained settings: containerised pipelines, transfer learning from publicly hosted weights, and datasets sourced where possible from the host country's own collections.
+
+The first workshop I would scope sits with the H3ABioNet / H3Africa community, who have spent over a decade building African genomic infrastructure and have been systematically underserved by the foundation-model wave.
+Year-two and year-three editions expand to South-East Asia (probably anchored at the MORU collaboration in Bangkok or NUS in Singapore) and Latin America.
+Each edition leaves behind a containerised teaching environment that runs on modest hardware, a small number of fine-tuned community models on locally-relevant tasks, and a peer cohort of trainers who can run later editions independently.
+I would budget the fund for travel cost-sharing for participants who cannot self-fund, because that is where most "open" programmes silently fail.
+
+Beyond the workshops I would use the ambassadorial role to push for honest evaluation practices in published genomics-AI work, specifically benchmarks that include ancestry- and tissue-diversity stratifications, in line with the Grand Challenge framing above.
+I have done versions of this community-shaping work already: founding the Virtual Cell Journal Club at Valence, contributing to napari and the Bioimage Model Zoo, and co-authoring the AI4LIFE €5M federated bioimage AI infrastructure grant.
+
+## In closing
+
+I am a research engineer.
+I came up through instrument-building, went into industry to ship multi-modal foundation models at scale, and want to come back and do this work in the open.
+Sanger is the one place in the UK where the data, the strategic commitment to AI, and the global-health framing live in one building.
+I want to spend the next three years there.
+
+Sincerely,
+Craig T. Russell, PhD