Leading openJII: product, teams and practical AI

Role: Technical Program Manager, openJII · Organisation: Jan IngenHousz Institute · Period: March 2026–present

The challenge

Plant scientists work across instruments, field sites and disciplines. Collecting measurements is one part of the job. They also need to understand their data, apply analysis methods and collaborate with other researchers.

At the Jan IngenHousz Institute, openJII brings those activities into a shared, open-source platform. That creates a leadership challenge as well as an engineering one: deciding which research needs should become reusable product capabilities, coordinating internal specialists and external developers, and making the resulting tools useful in everyday work.

My responsibility

I own openJII's technical vision and roadmap and lead its internal and external development teams. I work with scientific leads to translate research priorities into product and engineering decisions, while connecting the platform to the institute's wider programmes.

My remit spans architecture, delivery, hiring and capability-building, cloud costs and technical coordination with development partners. I help define requirements and recommend where to build internal capability or work with a partner. Commercial contract approval remains with the institute's director.

The platform is a team achievement. My contribution is to set direction, make the trade-offs explicit and connect the people doing the work. I stay close enough to implementation to test assumptions and help resolve difficult problems.

Helping existing research continue

The September board snapshot recorded 205 experiment IDs transferred through the PhotosynQ import stream. Bringing historical experiments into openJII preserves existing research work and gives teams a way to continue on the new platform. In one assisted workflow, we brought two historical trials into openJII, connected a scientist's analysis to the data and returned results to dashboards for inspection. My role is to connect migration capability with the workflows and support needed for that transition.

Connecting product decisions to evidence

One small mobile-interface decision illustrates the approach. A measurement workflow assumed that a researcher's important identifier would contain the word “plot.” But an analysis I had run across 45,451 research-question rows from 19,658 historical PhotosynQ projects showed that researchers more often used treatment, replicate, plant, leaf or genotype identifiers.

I brought that evidence into the review. It gave the team a concrete basis for reconsidering the assumption and supporting a wider range of research workflows. This is how I approach product direction: understand the users, make the trade-off visible and use evidence to decide what should change.

Read the product-decision example.

Putting AI to work

AI is part of both the organisational work and my own delivery method. I have developed programme proposals covering staff skills, AI-assisted development, tooling, governance and research collaboration. Those decisions matter because adopting a tool does not, on its own, give a team a reliable way to use it.

In my day-to-day work, I guide AI assistants through research, analysis, prototyping and engineering, then review the outputs against the task and its evidence. In one research-support workflow, an initial exploration appeared to disagree with a manuscript figure. Clarification with the scientist established that the figure used a separate analysis procedure, rather than the platform's existing calculation.

I worked from that procedure to package a reproducible analysis, connect it to platform data and return results to inspectable dashboards. All six checked group counts matched the figure. The scientist supplied the method and experimental context; my contribution was directing the packaging, integration and verification work.

AI extends the work I can carry through myself while remaining responsible for product and programme decisions. That experience also helps me judge what an organisation needs around AI: clear ownership, usable data, staff capability and a way to check whether the result is fit for purpose.

Choosing where to build capacity

I co-developed an engineering organisation proposal that compared internal hiring with specialist support and made the consequences of constrained funding explicit. The question was which capabilities needed an internal owner, what could be commissioned externally and which dependencies would remain if investment was deferred. The proposal set out alternatives for leadership rather than assuming every role could be funded.

A platform in use

On 15 October 2026, openJII's public metrics reported 48,026,412 measurement records in production. Of those, 40.5 million had measurement timestamps in the preceding 30 days, across 86 experiments. The development environment held a further 37.6 million records, taking the combined environment count above 85 million.

The production and development datasets are separate. Together they show the scale of the platform I help lead; the production figure gives a clear view of the live environment. These are measurement-record counts, not counts of users.

The work also extends into helping researchers use the platform. I served as a trainer and mentor at the March 2026 Photosynthesis Hackathon in Accra, where openJII supplied curated datasets and shared analysis tools. I am a coauthor of the resulting bioRxiv preprint. Alongside delivery, I write about the platform's architecture, research workflows and product changes.

What I bring to a fractional CTO engagement

This work brings together the responsibilities I help founders and leadership teams with: deciding what to build, shaping the team and partnerships needed to deliver it, and carrying technology decisions into working products.

My role is to connect the business objective to the people, operating practices and technical work it requires. I am completing an MBA in AI, Data & Analytics at the University of Amsterdam alongside this work. The staffing choices and research workflows above show the questions I work through in practice: where scarce capacity should go, what the team needs to own and whether a result is useful enough to carry forward. Practical AI experience lets me help assess an opportunity, test it and support the team that will own it.

Explore the work

Measurement figures are snapshots from 15 October 2026, with separate production and development datasets. Migration figures refer to the September board snapshot.