◢ Deep Tech R&D Boutique
Advanced Deep Tech &
Signal Processing
We design custom sensing systems, advanced DSP pipelines, and edge-AI hardware — and build our own AI-powered GPR robot for humanitarian demining.
— Flagship Product
AI-Powered GPR Survey Robot for Humanitarian Demining
An autonomous ground robot that surveys terrain with ultra-wideband ground-penetrating radar and classifies buried explosive hazards — including plastic-cased mines that metal detectors and drones miss. Our AI is trained entirely on FDTD physics simulations, eliminating the need for real minefield data. The robot identifies and prioritizes threats, feeding precise data to mechanical clearance assets.
— Industry Applications
Commercial R&D delivered
We partner with industry leaders to design custom signal-processing pipelines, deploy production-grade ML, and translate research into shipping products.
Client / Clinical Devices
Case StudyDX-Systems
Veylantis developed a deep learning–based filtering pipeline for biomedical signals in electrodiagnostic equipment. Designed and validated a signal denoising module for the manufacturer's companion software running alongside the diagnostic hardware.
Client / Embedded Sensing · Sweden
Case StudyStrikersoft
Veylantis founder Mike Schaeffer contributed to R&D in mmWave/FMCW radar signal processing and edge AI for sensing applications. Developed real-time DSP components optimized for resource-constrained embedded platforms.
— Open Source & R&D
Selected portfolio
A growing collection of research projects, simulation frameworks, and contributions to the open-source signal-processing community.
FMCW Radar Simulation
3D raytracing engine with Bartlett beamforming for FMCW radar systems. Simulates range-doppler-angle response in realistic multi-target environments.
UWB Radar Classification
Feasibility study on single-A-scan, single-position UWB shape classification. Benchmarks raw-signal, handcrafted-feature, and hybrid deep learning pipelines on FDTD-simulated data across 5 canonical shapes — reaching 91% accuracy in the minimal-aperture regime.
Neuromuscular EMG Simulation (MyoGen)
Open-source contributions to the neuromuscular simulation framework. PR #10: fixed unit mismatch in fiber radial distance. PR #11: added optional GPU acceleration for SFAP computation via CuPy.
— Publications & Research
Writing & technical articles
HotThe Number e and the Logarithm — Explained Intuitively
Why every loss function in machine learning seems to involve a logarithm — and where this strange function actually comes from.
Shannon Entropy Demystified: The Intuition Behind ML's Most Important Metric
How Claude Shannon turned probability into the universal language of compression — and why every neural network speaks it.
NewCross-Entropy Demystified: The Price of a Wrong Model — and How Half of AI Learns From It
Why cross-entropy sits at the heart of nearly every modern classifier — and what it really measures.
— Why Clients Work With Us
Engineering-Focused Collaboration
— Let's Build Together
Building a Radar, DSP, or Embedded AI System?
Whether it's a custom sensing system, a production ML pipeline, or a full hardware prototype — let's discuss your technical challenge and map the path from research to a shipping product.
— Workflow
How We Work
Initial Consultation
We begin with a technical discussion focused on your system requirements, constraints, timelines, and engineering objectives.
System Analysis & Architecture
Our team evaluates feasibility, signal chains, hardware constraints, data pipelines, and deployment requirements to define a scalable technical architecture.
Research & Prototype Development
We develop simulation environments, DSP pipelines, radar processing modules, embedded AI components, and proof-of-concept implementations tailored to the project goals.
Optimization & Integration
The system is optimized for real-world constraints including latency, computational efficiency, embedded deployment, and RF performance.
Validation & Testing
We perform benchmarking, signal validation, robustness testing, and performance evaluation using both simulated and real-world datasets where applicable.
Deployment Support
We assist with deployment, integration, iterative improvements for production.
— In-House Infrastructure
Real lab. Real hardware.
We design, prototype, and manufacture in-house — from custom enclosures to field-tested robotic platforms.
3D Print Farm
Rapid prototyping at production scale
A farm of 13 Bambu Lab printers (X1C / P1S / A1) lets us iterate on custom enclosures, mounts, and mechanical parts in hours, not weeks — and produce small batches in-house without outsourcing.
3D Print Farm
13 Bambu Lab printers for rapid prototyping and small-batch production.
4WD Robotic Platform
Heavy chassis with 80 kg payload, spring suspension, field-tested off-road.
Embedded & Sensing
Xilinx Zynq / PYNQ-Z2 FPGA boards and UWB transceivers for real-time edge inference.
— Team
The people behind Veylantis
Alina Tiutiunnyk
CEO
Strategy, partnerships, and quality assurance. Leads communication, process, and international collaboration.
Mike Schaeffer
Co-founder, Tech & ML Lead
Signal processing, machine learning, radar DSP, and biophysical simulation. Translates physics into production-grade AI.
Bogdan Tishchenko
Co-founder, Hardware Lead
Robotics, electronics, and additive manufacturing. Builds and field-tests the physical platforms.
— Contact
Tell us about your project
Share your technical challenge and we'll get back within one business day.