Available for contract · London / Remote

Si Hammond

AI systems for messy real-world problems

I design, build, and ship systems that help people make better decisions with complex information — from clinical AI and labour-market intelligence to human-centred AI products and full-stack prototypes.

What I bring

Applied AI & GenAI

I build AI systems that go beyond demos: retrieval and search workflows, structured extraction, evaluation loops, and product logic that can survive contact with real users.

Clinical & Health AI

I’ve worked inside a clinical AI startup from prototype through to paid trials, and understand the gap between a persuasive demo and a system a clinician can actually trust.

Full-Stack & Prototyping

I can take an AI idea from architecture through to a working web application. Python backends, modern JS frontends, data pipelines, knowledge graphs — I've built and shipped all of it, often as a team of one.

Building systems that make complexity tractable

My PhD in Evolutionary Computation gave me a grounding in optimisation and search that turns out to be surprisingly useful when building AI systems that need to do more than generate plausible text. I've spent the last fifteen years moving between data science, full-stack development, and — more recently — the fast-moving world of generative AI and agentic systems. I've worked in pre-seed startups, post-acquisition businesses, and everything in between.

Experience

  • Founder, ShellSi LtdCurrent
  • Machine MedicineCTO & Co-Founder
  • WorkDigitalChief Data Scientist
  • VONQ · Recruiting BrainfoodData Science & Full-Stack

Education

  • PhD, Evolutionary ComputationUniversity of Birmingham
  • MSc, Natural ComputationUniversity of Birmingham
  • MSc, Evolutionary & Adaptive SystemsUniversity of Sussex
  • BSc, AI & Computer ScienceUniversity of Birmingham

Where I work

01

Health & Clinical AI

Healthcare is where the stakes are real. I'm drawn to problems where AI can reduce clinician burden, surface insights from messy clinical data, or help patients navigate complex systems — and where the regulatory and safety constraints make the engineering genuinely hard.

02

Education & EdTech

Education sits at the intersection of two things AI is genuinely good at: personalisation at scale, and making sense of complex unstructured data. There's significant work to be done here, and I'm interested in doing it.

03

Labour Market Intelligence

I've spent years building tools that make sense of job markets, skills data, and career trajectories. It's a domain I know deeply — and one where AI is creating both new possibilities and new responsibilities.

Selected work

View all selected work →

What people say

Simon possessed the intellect and creative problem solving abilities required to develop some of the key methods that we use to interrogate/structure/package these datasets into commercial products that we sell globally. He also was able to take very complex technology and make it understandable to a wider group within the company.

Bill Fischer CTO at VONQ · Board Advisor · Founder

Simon is a conscientious and hard working developer, he is extremely analytical with a keen eye for detail and an ability to grasp complicated concepts with little effort. He is able to work equally alone as with a team and is always active, attentive and productive when working with others. He is very well organised and uses this to effectively focus on the task at hand while still being cognisant of the rest of his work load.

Bahul Upadhyaya Innovator, architect, and developer

Simon is a very high-level thinker, bringing about high-level concepts to reality through his artisan coding abilities.

Sanford Dickert Director, Product / Technology @ Kindred Factors · CTO · Director of Product

Bringing Simon in brought immediate results. Our algorithms weren’t quite right. Simon quickly picked up on both the business logic and how users would behave and made quick headway toward a solution that was elegant and quick to implement.

Ray Rafiq Unretiring… · Client

How I work

The pattern is usually the same: reduce ambiguity, test the risky bit early, make the system legible, then keep iterating until it becomes genuinely useful.

01

Clarify the problem

Get specific about the user, the decision, and the constraint that actually matter.

02

Prototype fast

Test the risky part early rather than debating it in the abstract.

03

Make the system observable

Add the logging, evals, and feedback loops needed to see what it is really doing.

04

Iterate toward something usable

Turn the prototype into a tool people can trust, understand, and adopt.

Let's talk

I’m available for contract work on AI products, decision-support tools, and other systems that help people make better use of complex information. Based in London and working remotely, I’m particularly interested in healthcare, education, and career-related problems — but I’m always open to the rare knotty problem with real stakes.

si@shellsi.com LinkedIn → linkedin.com/in/sihammond