A scoping call to define the problem.

The ones where the hard part is the system, not the headcount. Compilers and optimised backends, GPU and HPC workloads, formal verification and microkernels, agentic control loops, and on-prem or air-gapped ML. If the problem gets solved by adding engineers, you probably don't need us.

With a scoping call to define the problem, then a short fixed engagement that ends in a written problem statement and a plan. That plan is yours either way, including if you decide to build it with your own team.

Both founders, across the whole engagement. Where a problem needs a domain specialist in verification, hardware or quantum control, we bring one in and name them to you before they start.

You own all deliverables and their rights. Pre-existing and general-purpose tooling stays ours. Naming and publication are agreed before work starts, which is why the references on this site describe the work but not the client.

Every milestone produces something that runs. Verification produces numbers you can check yourself: tests, benchmarks or proofs, depending on what the work calls for.

The company is new, the work isn't. Between the founders there is over a decade across CERN-HSF, LBNL, Høgskolen i Molde, Appwrite and Homebase, and we are an seL4 tech partner. Current engagements include a Norwegian research laboratory, a German university machine learning lab and a US HR technology platform.