Selected work / Pharmatalent (Switzerland / UAE)
Pharmatalent
A multi-sided B2B SaaS talent platform with dual-algorithm ML search for regulated life-sciences.
Gave recruiters an ML search they could read and steer in their own words, instead of a ranking they had to trust.
- Dual-algorithm ML search with natural-language recruiter feedback
At a glance
- Client
- Pharmatalent (Switzerland / UAE)
- Sector
- Life-sciences · B2B SaaS
- Period
- 2025
- Product
- Multi-tier talent-advisory platform
- Role
- Lead product designer · Interface architect · Client lead
- Team
- Design + ML engineering
- Platforms
- Web app
Confidentiality Client-approved summary. Screens containing identifiable contact details are omitted; the images show the product's brand, login and an ML concept illustration.
What it was
A B2B SaaS talent-advisory platform used by three different groups — recruiters, candidates, and administrators — inside high-compliance European life sciences. The real job was to make complex, ML-driven matching feel simple and trustworthy to a professional recruiter.
The hard part
Three audiences who each wanted different things, a regulated setting that raised the bar on trust and data handling, and a matching engine whose whole value hinged on one thing: whether recruiters believed its results and could push back on them. If they couldn’t, the cleverness underneath was worth nothing.
What I did
I led the client relationship and the product-design work from the first framing through to delivery. That meant advising the senior stakeholders on what to prioritise, and turning the ML story into journeys, interaction rules, and an interface a recruiter could actually understand and control.
The decision that mattered most
We made feedback a first-class part of the flow. Recruiters could correct the results in their own words, right inside the search, so the dual-algorithm matching learned from their expert judgement instead of handing them a ranked list and expecting them to take it.
Where it landed
Recruiters stopped arguing with a ranked list and started correcting it, in their own words, and the matching got better as they did. Finding the right people got a lot less painful in a corner of European healthcare where the rules leave little room for guesswork.
