Advisory
Bryan Tegomoh, MD, MPH advises the teams making consequential decisions about AI in the life sciences: biotechnology and pharmaceutical companies, the investors and frontier labs backing them, and research institutions evaluating AI systems for biomedical discovery and translational research. Engagements are independent and evidence-centered, with emphasis on model evaluation, biological validity, reproducibility, and responsible research practice.
Advisory Scope
- Model and dataset review. Independent assessment of AI systems for protein design, genomics, single-cell, imaging, or therapeutic discovery workflows. Deliverable: written assessment with capability tier (Demonstrated, Theoretical, Beyond Current Capabilities), failure-mode catalogue, and validation recommendations.
- Evidence review for AI claims. Triage of vendor pitches, paper claims, and competitor analysis. Deliverable: written review with credibility-hierarchy placement and citation-verified evidence assessment.
- Benchmark design and validation strategy. Design of biology-aware splits, validity checks, and prospective validation plans for internal AI programs. Deliverable: validation plan with split design, success criteria, and stop rules.
- Research governance and reproducibility. Internal frameworks for DOME-compliant reporting, model and dataset cards, dual-use review, and release decisions. Deliverable: governance policy and implementation roadmap.
- Information hazard review. Pre-publication or pre-release review of capability work that could lower barriers to misuse. Deliverable: written assessment with publication-versus-controlled-release recommendation.
- Workshops and team briefings. Half-day to multi-day workshops for research teams adopting AI in biomedical discovery. Deliverable: briefing materials, evaluation framework, and reading list.
How Engagements Work
Initial scoping is brief and non-binding. A short call or email exchange establishes the decision context, the deliverable, the timeline, and any conflict-of-interest considerations. Engagements proceed when scope fits available time and conflicts are clear.
Deliverables are written. Verbal-only engagements are not the default because the value of independent advisory work is in the documented assessment that the client can later cite, reproduce, and act on. Written deliverables also create a record if an assessment is later reviewed.
Citation discipline matches the handbook standard. Every claim in an advisory deliverable is anchored to a primary source where one exists. Vendor-reported metrics are labeled. Preprints are labeled. Where evidence is mixed, the lower-confidence tier is preferred.
Independence and Conflicts
Advisory work is independent. It does not imply endorsement of any product, organisation, or public claim. Recommendations are based on peer-reviewed evidence and reproducible benchmark results, not on commercial relationships.
Potential conflicts of interest are disclosed at the start of an engagement. Concurrent advisory work with directly competing organisations on the same topic is declined. Equity or option arrangements that would compromise independence are not accepted.
Topics Out of Scope
- Patient-specific medical advice or clinical decision-making.
- Operational biosafety consulting (BSL operations, select-agent handling).
- Step-by-step operational details that would lower barriers to misuse.
- Product endorsement, paid public-facing testimonials, or marketing collaboration.
- Legal or regulatory determinations (engage qualified counsel).
Inquiries
Tell me the organisation, the decision in front of you, the scope and timeline, and any materials worth reading first. Initial scoping is non-binding.