Data Swamp Biosystems  ·  Cell Painting AnnData Validator  ·  Bio Run Crate  ·  BioDataset Scout

Building trustworthy scientific data systems

I work across scientific research, data strategy and software engineering. My focus is making complex research data discoverable, reproducible, traceable and ready for responsible AI/ML use.

Scientific data strategy Target operating models, platform architecture, data products, metadata strategy, stewardship and governance-as-code.	Computational biology Genomics, single-cell and spatial omics, bioimaging, phenotypic profiling and translational research data.
Software and data engineering Typed Python, validation tooling, Nextflow, containers, cloud platforms, CI/CD and reproducible scientific workflows.	Programme delivery Technical roadmaps, architecture decisions, cross-functional delivery, platform evaluation and measurable governance outcomes.

Flagship open-source systems

Data Swamp Biosystems A deterministic synthetic biotech data estate for benchmarking catalogues, lineage systems, governance controls and AI agents. synthetic-data data-governance AI agents	Cell Painting AnnData Validator Semantic, provenance and AI-readiness validation for Cell Painting datasets represented as AnnData. anndata phenotypic-profiling validation
Bio Run Crate Validation of biological analysis-run metadata and packaging of traceable outputs as RO-Crate research objects. RO-Crate FAIR reproducibility	BioDataset Scout Evidence-backed discovery, qualification and acquisition intelligence for biomedical datasets, with explainable scoring and preserved provenance. biomedical-data data-discovery provenance

Open-source portfolio pulse

_{The two panels above are generated as local SVG files by GitHub Actions. The profile does not depend on a public badge or statistics-image service.}

Engineering principles

• Contracts before convenience: make schemas, interfaces and expectations explicit.
• Reproducibility by default: deterministic execution, versioned configuration and traceable outputs.
• Governance in the workflow: validation, lineage, ownership and quality controls belong in the delivery path.
• Build for scientific users: technically rigorous systems must remain understandable and usable by research teams.

Technical landscape

Scientific computing
Python · R · AnnData · Scanpy · Squidpy · Cell Painting · single-cell omics · spatial transcriptomics · bioimaging

Engineering
Nextflow · Docker · Kubernetes · GitHub Actions · pytest · Ruff · mypy · Pydantic · uv

Data platforms
AWS · Google Cloud · S3 · DataHub · OpenMetadata · OpenLineage · Benchling · Seqera

Strategy and governance
FAIR · ALCOA++ · metadata management · data quality · data lineage · data products · stewardship · governance as code

---

Scientific data should be understandable, reproducible and trustworthy by design.
London, United Kingdom