01 Where I can contribute
Start with the constraint.
Then choose the intervention.
Not every problem needs an agent. Sometimes the right answer is a review, a deterministic
rule, a better integration, or deciding not to add a model at all.
AI Systems & Workflow Automation
Where does work get stuck, and what should never be left to a model alone?
Map a real workflow, identify where retrieval, model reasoning, deterministic checks,
integrations, or approval fit, then design the smallest useful system around it.
USEFUL OUTPUT
A workflow map, system boundaries, evaluation plan, and—if in scope—a working slice with a clear human handoff.
A fit for: teams automating research, document work, internal operations, or multi-step processes.
02
ARCHITECTURE & RELIABILITY
Technical Architecture & System Reviews
What breaks first—and what is worth fixing before adding more?
Examine a proposed design or existing system in the context of its constraints:
data flows, failure modes, integration seams, access controls, operational ownership,
and the role AI should actually play.
USEFUL OUTPUT
A prioritised technical review: observed evidence, material risks, trade-offs, open questions, and practical next decisions.
A fit for: pre-build reviews, production readiness, a difficult architecture choice, or a system that needs a second opinion.
AI Product Engineering & Implementation
How should this capability fit into the product you already have?
Build a scoped AI feature or backend integration within the existing product
architecture—not a disconnected demo. Make the seams, error cases, evaluation, and
operating assumptions explicit as the feature takes shape.
USEFUL OUTPUT
Agreed implementation, integration code, proportionate tests, documented decisions, and a maintainable handover.
A fit for: shipping a first AI capability, integrating model services, or turning a useful prototype into product work.
Technical Due Diligence & Fractional Technical Leadership
What does the evidence support—and what still needs to be proven?
Review a product, codebase, system design, or technical plan against the claims and
context that matter. Or work alongside a team on engineering priorities, architecture,
and technical decisions within an agreed remit.
USEFUL OUTPUT
For reviews: evidence-linked findings and follow-up questions. For leadership: agreed decision support and hands-on ownership boundaries.
A fit for: technical evaluations, early product decisions, or teams that need senior engineering input for a defined period.