Research & development
Applied research and machine learning for questions without an off-the-shelf answer. We deliver reproducible prototypes and results that hold up to review.
Turn hard engineering questions into working systems.
Peinser is a Belgian engineering team for research, platform engineering, and aerial robotics. We find the right approach, prove it, and put it into production.
Clients include Acerta, ULiège, Innovatix, Duvel Logistics, H.Essers, and Van Cleven NV.
Platform engineering
Distributed systems, Kubernetes, CI/CD, and observability for teams whose platform slows delivery. We design the target, ship the first production slice, and hand it over.
Aerial robotics
Mission planning, ground control, payload integration, and autonomy for custom fixed-wing, VTOL, and multirotor UAVs.
Strategy
Technical audits and decision memos before an expensive or hard-to-reverse choice. We make the options, costs, and risks explicit before you commit.
Products
Flightdeck
Mission planning and ground control for custom UAVs. Plan against aircraft performance, forecast wind, terrain, airspace, and radio coverage, then fly the mission from the same cockpit. In alpha.
Xarta
Render, sign, archive, and dispatch documents through one deterministic pipeline, from versioned templates to PDF/A and Peppol invoices. On-premises by default.
Ask about Xarta →Zonnetanken
EV charging that follows the sun, with auditable sessions for employee cost recovery. An independent product with paying users across Belgium.
Visit Zonnetanken →Start with six weeks.
Long enough to settle the important question. Short enough to act on the answer.
Discover
Weeks 1 and 2. We read the code, the data, the papers, and the runbooks, and talk to the people who run the system. You get a findings memo with every assumption stated.
Decide
Weeks 3 and 4. A short list of options, each with a cost, a risk, and an owner. We choose together and write down why.
Ship
Weeks 5 and 6. The decision meets real users, real infrastructure, or a representative flight. You get a working, instrumented first slice and a plan for what comes next.
Open source
peinser/corshub
Federated network of RTK base stations broadcasting RTCM corrections.
peinser/quic-sni-router
Routes QUIC traffic to Kubernetes pods by SNI, without terminating TLS.
peinser/esp-remoteid
EU Direct Remote ID broadcast firmware for the ESP32-S3.
JoeriHermans/amortized-experimental-design
Tutorial and code for learning to select optimal experiments.
Research
Likelihood-free MCMC with amortized approximate ratio estimators
ICML 2020
Towards reliable simulation-based inference with balanced neural ratio estimation
NeurIPS 2022
A crisis in simulation-based inference? Beware, your posterior approximations can be unfaithful
TMLR 2022
Constraining warm dark matter with stellar streams and neural networks
MNRAS 2022
Bring us the problem before the specification.
Tell us what is uncertain, what you have tried, and what a good outcome would change. We are booking engagements from Q2 2027.