AI Product Owner
Josen
I build AI and data products end to end — from the messy problem to the shipped, running thing. TopListers, the product you are on, is the clearest example.
Ten years as a Senior Product Owner and Business Analyst, now building AI and data products hands-on. I own the whole loop — framing the problem, making the product calls, and staying close enough to the engineering to ship and keep things running.
Flagship case study
TopListers — job-market intelligence, mapped to the world
- 100k+
- live roles, verified daily
- Dozens
- of countries covered
- Direct
- from-employer ATS sourcing
- 0
- ads — free to post
The problem
Job data is fragmented and region-blind, and aggregators bury direct-from-employer roles behind ads and duplicates. Worse, most boards answer the wrong question — “does this job exist?” — when the one that decides whether a role is real for you is “can I actually work there?”
The product decisions
Source direct from employers, not aggregators
Pull straight from employer ATS feeds (Greenhouse, Lever, Ashby) instead of paid aggregator APIs — fresher listings, de-duplicated, each one linking back to the original posting.
Location-first discovery
A 3D globe is the front door, so you can see where the opportunity actually is before you start filtering — not a wall of listings with a search box bolted on.
Answer “can I work there?”, not just “does the job exist?”
Visa-pathway intelligence and a live sponsor register sit alongside every listing, so the product answers the question that actually decides whether a role is real for you.
Free, ad-free, terms-respecting
No ads, every listing free to view, and if a source’s terms change we pull it. Trust is the product; the sourcing pipeline is built to keep it.
Under the hood
Next.js (App Router) · PostgreSQL + Prisma · BullMQ / Redis worker queues · a scheduled fetch → normalise → geocode → dedupe → upsert sourcing pipeline across direct employer ATS feeds and open providers · a d3-geo / TopoJSON globe · a public JSON API with keys · email digests · first-party analytics.
Why it matters as an AI-PO artifact
It is a real product with real users and a real data pipeline behind it — not a slide deck. It shows product strategy and prioritisation, and enough technical depth to make the architecture calls and keep the thing running in production.
What I build
The product you are on. Globally-aware job-market intelligence with a 3D-globe discovery interface, direct-from-employer sourcing, and work-authorisation intelligence.
An AI WhatsApp assistant that turns an inbound message into an instant, deterministic quote and a booked job — replacing manual phone-and-diary handling for home-services businesses.
An AI-agent platform that automates customer service, scheduling and back-office operations for UK service businesses. Deployed first inside a live operation, productising toward a SaaS launch.
An F1 Pit-Stop Optimiser and a Climate × Stock-Market intelligence model — predictive models built from MSc dissertation research.
How I work
Product ownership with business-analysis rigour, plus hands-on fluency with LLMs, data pipelines and the systems around them. MSc in Data Science & Computational Intelligence (Coventry, merit); SAFe-certified Product Owner / Product Manager.