Anton Kornilevsky – Freelance Embedded ML (TinyML), Federated Learning & AIoT Engineer

About & focus

I am an embedded AI engineer with 10+ years of experience in Embedded Linux, bare-metal systems and firmware verification. Through my company NeuraNidus UG I help teams bring machine learning onto resource-constrained devices – from microcontrollers to industrial gateways – while keeping raw data on the device.

Embedded Machine Learning (TinyML)
Federated Learning
AIoT – Artificial Intelligence of Things
Real-time inference on MCUs, NPUs & DSPs
Embedded Linux & IoT gateways

Skills

Languages: Embedded C, C++ (C++17), Python
Embedded ML / TinyML: TensorFlow Lite, PyTorch, ONNX, YOLO, quantization flows, federated learning, STM32 Nucleo-N657X0-Q
Operating systems & embedded software: Embedded Linux (Yocto, Buildroot), RTOS, bare-metal, POSIX
Hardware platforms: ARM Cortex-M (STM32N6), ARM Cortex-A8, ARMv7 / NXP i.MX6 UltraLite, ARM926 (Microchip SAM9), BeagleBone Black, Infineon AURIX TriCore TC3xx, NXP QorIQ, Freescale MPC56xx, Intel MAX 10, NPUs, DSPs
Compiler engineering: Optimization passes, IR lowering, backend mappings, code generation
IoT, cloud & protocols: Azure IoT Hub, Azure Machine Learning, Azure OpenAI, MQTT, REST, TCP/IP, SPI, I²C, CAN bus, LVDS
CI/CD, testing & tooling: CI/CD with Jenkins and GitLab CI, Docker, CMake, Git, unit testing (Google Mock, CMock), GDB, Lauterbach, UDE, Doxygen
Standards & domains: Automotive (ISO 26262), avionics hardware (DO-178 / DO-254), cellular/IoT modem firmware verification
Spoken languages: English, German, Hebrew, Russian (all fluent)

Selected projects & experience

Nidal.io – privacy-preserving federated edge AI platform

Building a platform to train, deploy and scale models from MCUs to industrial gateways without raw data ever leaving the device.

Embedded Linux on a large IoT gateway fleet

Developed embedded Linux applications (C, ARMv7 / i.MX6) rolled out to more than 150,000 IoT gateways; cut IoT data traffic by 70% by optimizing Azure IoT communication, with over 85% test coverage in component and end-to-end tests.

Automotive radar verification (ISO 26262)

Developed C-based verification libraries and test cases for Infineon AURIX TriCore radar chipsets – bare-metal, inline assembly, SPI/I²C/LVDS – with Jenkins-based test automation.

Embedded Linux systems & middleware

Built and maintained Linux-based embedded systems with Buildroot, designed REST APIs and middleware, and containerized builds with Docker on ARM Cortex-A8 / ARM926 targets.

Modem firmware verification & hardware design

Validated cellular and IoT modem firmware and extended a Python/Qt verification tool; earlier reviewed hardware designs and test plans for DO-178 / DO-254 avionics systems.

Availability & engagement

Available for freelance projects via NeuraNidus UG for up to 2 days per week, remote or on site in Germany.

Rates on request.

Education & training

Embedded Linux development – certificate, Doulos
Embedded machine learning (seminar) and foundations of AI (course) – Universität Duisburg-Essen
C++ primer for embedded systems – Digi-Key Continuing Education Center

Kurzprofil (Deutsch)

Anton Kornilevsky ist freiberuflicher Embedded-AI-Entwickler und Gründer der NeuraNidus UG (haftungsbeschränkt) in Duisburg (NRW). Mit über 10 Jahren Erfahrung in Embedded Linux, Bare-Metal-Systemen und Verifikation unterstützt er Unternehmen bei Embedded Machine Learning (TinyML), Federated Learning und AIoT – von Mikrocontrollern bis zu industriellen Gateways. Verfügbar für bis zu 2 Tage pro Woche, remote oder vor Ort.

Contact

For project inquiries, please email me directly:

anton.kornilevsky@neuranidus.com

Company website: www.neuranidus.com

Note: I only send project communication from my @neuranidus.com address. If someone contacts you on my behalf from another address, please verify by emailing me directly.