I design systems,build products, and makecomplicated technologyunderstandable.

Turning lived experience into a readable mapping

Problem

Explicitation interviews worked. The technique gets someone to describe what they actually did, not what they think they should say, but it needed a trained expert in the room. One interview ran an hour to ninety minutes, and turning it into a formal mapping of someone's competences took another five to six hours of expert work afterwards. That put the method out of reach of the people who needed it most: career changers, job seekers, and the counsellors trying to help them at scale.

Role

CTO & lead engineer

Decision

Train a specialised classifier instead of prompting a frontier model, after proving the frontier models plateaued

Outcome

An interview anyone can take, in 15 minutes to an hour instead of 90, producing a mapping in five to six minutes instead of five to six hours. Across three rounds of testing with 36 participants, every respondent agreed the interview left them enough time to think, and the people with no prior chatbot experience gave it the highest recommendation of anyone. It all feeds two products built from one codebase: yepa.solutions for individuals, and yepa.expert, invitation-only for researchers, with code-gated campaigns, anonymous respondents and group-level analysis.

Forecasting a bin so trucks stop driving to empty ones

Problem

KOCO's existing forecast was an adaptive linear regression: average the fill rate between two collections and extrapolate. It works, but it only ever sees data from collection days, a handful of points a week. The company wanted to know whether eighteen months of sparse weighings could support something better.

Role

Research engineer

Decision

Lower the resolution from hourly to daily, because at hourly, 98% of the data was a line I had drawn myself

Outcome

Prophet cut mean absolute error from the incumbent's 75.0 kg to 48.04 kg on real collection days, and MAPE from 40.4% to 19.8%. Both partners rated the project 1 of 4 (very successful) in the final report to Innosuisse.

Reading financial news three questions at a time

Problem

Off-the-shelf financial sentiment tells you a headline is negative. It rarely tells you negative about whom, or whether the thing described has already happened. A single article can be optimistic about one company and grim about another, and can mix an event from last year with a forecast for next quarter. Treating that as one score throws away most of the signal.

Role

Sole author

Decision

Put the temporal class boundaries in a configuration file rather than in the model's weights

Outcome

Graded 6 out of 6. Three separable questions answered per sentence fragment, aggregated per company and per time window, with named emotion pairs instead of a forced single winner.

A robot you can drive from your sofa

Problem

The idea came from a florist. Adrien Hertig noticed his customers had stopped coming into town and were ordering online instead, and thought it would be good if they could still be walked around the shop by one of his staff. Then COVID closed the shops entirely. The question became whether a remote customer could genuinely browse a real shop from a phone: see what they wanted to see, be served by a person.

Role

Lead developer, technical lead

Decision

Give up on peer-to-peer and make our own server a real WebRTC peer on both sides, because we had to see the pixels

Outcome

Ran in three partner shops in Fribourg (a florist, a furniture shop and a relaxation retailer) with a mobile app on both stores and real-time face blurring that let the robot operate in a shop with customers in it.