Mathijs Henquet
Software Engineer · Amsterdam, NL
About
I take ideas from pitch to production — identifying problems, building solutions, and driving them to adoption.
At Royal NLR, I consistently bridged the gap from research to operational impact. I pitched and built SKY, a secure generative AI platform for classified data now used organization-wide, delivering €800k+/yr in measured productivity gains. I then led SENTINEL, a NATO-funded voice assistant for military situational awareness, to adoption as a standard capability by the Dutch Ministry of Defence — a milestone reached by only 4% of defense innovation projects.
My work is grounded in a dual background in CS and mathematics, with work in formal methods and homotopy type theory. My engagement with AI safety began as a 2019 MIRI Fellow and remains central to my interests.
Experience
Drove innovation from concept to production. Combining technology scouting, strategic proposal writing, project leadership, and hands-on development.
- SKY — Pitched, designed, and built NLR's in-house generative AI platform for sensitive/classified data. Security-compliant, locally hosted. Generates €800k+/yr in measured productivity gains. Proposed spin-off strategy and consortium model for cross-institute adoption.
.NET C#, gRPC, Python, RAG, TensorRT-LLM, Docker, Rust
- SENTINEL — Pitched, led the project team, and was primary developer of a real-time voice-driven assistant for military situational awareness. Combines speech recognition, speech synthesis, and vision-language models into a responsive system for field operators. Adopted as a standard capability by the Dutch Ministry of Defence — one of only 4% of defense innovation projects to reach adoption.
Python, Whisper, WebRTC, vLLM, VLM/LLM, TTS (Kokoro)
- Mesh-as-a-Service — Developed automated 3D modeling concept using neural reconstruction. First-author paper was runner-up for Best Paper at I/ITSEC 2024 (Simulation subcommittee).
- Flight Debrief Study — Led strategic study on automated flight debriefing summarization. Published at ICCAS 2024.
- Supervised two Master's students on 3D reconstruction and drone path planning.
Supported Covid-19 scenario modeling for decision makers, focused on the global south. Built model visualization in the frontend.
Awards & Fellowships
- Runner-up NLR Team Innovation Award (2025) — SKY team, nominated among organization-wide finalists (among ~1000 employees)
- Runner-up Best Paper, Simulation subcommittee, I/ITSEC 2024 — for Mesh-as-a-Service
- MIRI Summer Fellowship (2019) — Academic retreat for mathematicians and programmers working on AI alignment
- Winner, μKP Programming Competition (2016) — Best programmer among ~280 second-year CS students, Utrecht University
Projects
Formal proof of correctness of automatic differentiation in the presence of recursion and iteration. Utrecht University, Mathematics.
On homotopy type theory and its relation to higher topos theory. Utrecht University.
Publications
Henquet, Bellucci, Amghane, et al. "Mesh-as-a-Service: Automated 3D Modelling fast as l-AI-ghtning." I/ITSEC 2024. Runner-up Best Paper, Simulation subcommittee.
Henquet, Bellucci. "AI-Assisted Debrief: Automated Flight Debriefing Summarization and Competency Assessment." ICCAS 2024.
Hellinga, Henquet, Bellucci. "SENTINEL: Supporting Military Situation Understanding Through Conversational Vision-Language AI." Submitted.
Education
Technical Skills
Languages
Dutch (native) · English (C2) · German (C1)
Contact
Based in Amsterdam, the Netherlands.