Photonic Knowledge: a mobile geoscience lab

The problem

Most geological mapping of rock core samples happens with no on-site analytical support — often in spreadsheets. Photonic Knowledge was building hyperspectral imaging systems to change that: instruments that identify minerals in rock the way our eyes tell a ripe banana from a green one, by reading the light it reflects. The goal was never to replace the geologist, but to make their work faster, more reliable, and easier to share.

The approach

I joined as the third employee, working as a signal specialist, and grew into directing R&D and the engineering team. Through five years I kept designing and developing the scientific software myself — image and signal processing, data visualization, algorithm development and experimental validation, in Python — while leading the roadmap, deployment planning, CI testing, documentation, and technical support.

The product was a mobile imaging laboratory in a trailer, deployed to remote and mostly offline sites. Our lab technicians and geologists did most of the live testing, and staying genuinely responsive to their feedback — with a clear roadmap and extensive training — was essential both to the product and to a resilient work culture when things went sideways. I also managed the IRAP grants that funded parts of the work.

What came of it

The lab could analyze 500 metres of core in about a week — work that takes an on-site geologist about four weeks, or an off-site laboratory about twelve — at roughly a tenth of the cost per metre. I filed a patent for the presentation of analytical results (CA2813913A1, 2012), prepared a second on the calibration and correction process, and developed a third innovation in the algorithmic identification. Photonic Knowledge received the Deloitte Technology Fast 50 award in 2012.

The company closed in 2014 when the commodity cycle turned — a lesson in how much a deep-tech company’s fate rides on its industry’s vitality. The engineering habits stayed with me.