Artur Andrzejczak

Principal Software Engineer: Embedded Linux, Kernel Drivers & Camera Systems
Łódź, Poland

• cv as pdf
• github
• linkedin
• google scholar
• orcid

Summary

Principal Software Engineer with 15+ years in embedded software and Linux kernel development, working on automotive camera and image-sensor drivers for in-cabin ADAS: driver monitoring (DMS) and occupant monitoring (OMS). Develops the V4L2 kernel camera drivers used in HARMAN's driver- and occupant-monitoring programs: sensor bring-up, GMSL2 SerDes and MIPI CSI-2 integration, board support (BSP) on NVIDIA Tegra/Jetson and Samsung Exynos Auto, diagnostics and power management, up to start of production. Also worked on embedded RTOS and bare-metal firmware, PCI kernel and userspace drivers, hardware design and debugging, plus lab and production test tooling. Uses coding agents (Claude Code, GitHub Copilot) in daily work.

Technical skills

Languages

C, C++, Python, Bash, Assembly

Kernel & embedded

Linux kernel & device drivers, board support packages (BSP), platform bring-up, device tree, kernel interrupts, DMA, I2C/SPI, sysfs, power management (suspend/resume, S2R), VFIO/UIO, PCI/PCIe (AER, SR-IOV PF2VF), Buildroot, JetPack / L4T, embedded RTOS (Contiki OS), bare-metal firmware, MCAL / AUTOSAR (automotive)

Camera & imaging

Image sensor drivers (OmniVision, STMicroelectronics), IR / RGB / RGB-NIR modes, RAW pixel formats, MIPI CSI-2 (C-PHY / D-PHY), V4L2 subdev / media-controller, GStreamer (v4l2src, nvarguscamerasrc) and FFmpeg for capture validation and frame conversion, GMSL2 SerDes (Maxim MAX9295/9296, MAX96717/96724, MAX96716; pixel and tunnel modes), sensor bring-up, sensor boot and firmware upload, ISP / auto-exposure (AE), driver diagnostics and fault injection (functional safety, ASIL-B sensor), multi-camera synchronization, Android camera framework (Qualcomm)

Platforms

NVIDIA Tegra / Jetson (AGX Xavier, Xavier NX, Nano), Samsung Exynos Auto, Renesas RH850, Qualcomm Snapdragon (MSM89xx), ARM, Atmel AVR, MSP430, x86-64

AI-assisted development

Claude Code, GitHub Copilot, ChatGPT, Microsoft 365 Copilot

Tools & methods

git (Gerrit, GitLab, GitHub, Bitbucket/Stash), Perforce, GCC, GDB, Make, Autotools, Google Test / gMock, Coverity, Klocwork, CodeSonar, Bullseye, Jenkins CI, Black Duck (FOSS / SCA scans), Saleae, Introspect, Rohde & Schwarz, CANoe, DaVinci Configurator/Developer, Green Hills MULTI, Altium Designer, Sparx Enterprise Architect (architecture review), ALM (requirements and tickets); static analysis, unit testing, Agile / Scrum

Professional experience

Principal Software Engineer · Jun 2022 – present

HARMAN International (Samsung) – Camera Team, Łódź

Automotive Engineering Services, ADAS / Intelligent Cockpit. Ready Care, Ferrari and Bentley programs: driver monitoring (DMS) and occupant monitoring (OMS) systems. Camera-driver development, debugging and platform integration on NVIDIA Tegra and Samsung Exynos Auto, and technical support for the teams that build on those drivers.

Camera drivers & features

• Developed the vb56g4a (DMS) and vb1940 (OMS/Bentley) V4L2 kernel camera drivers, together with GMSL2 link, NvM, LED and under-display-camera (UDC) drivers. Both run on NVIDIA Tegra and Samsung Exynos Auto and cover IR, RGB and RGB-NIR modes, runtime-detected camera variants, and interchangeable Maxim GMSL2 SerDes combinations (MAX9295/9296, MAX96717/96724, MAX96716) with MIPI CSI-2 (C-PHY/D-PHY) capture, pixel and tunnel modes, and sensor boot and firmware upload.
• Designed the driver's power model as a multi-level sleep and wake state machine exposed to userspace, and delivered multi-camera cold boot, reset recovery, adaptive 3/6 Gbps GMSL2 link, DMS/OMS frame synchronization and dual-camera parallel operation.
• Built the diagnostics subsystem (temperature, PoC voltage, link status, eye-opening monitor, imager faults) with fault injection, feeding the functional-safety (FuSa) diagnostics of an ASIL-B image sensor; added NvM readout of camera identity and calibration data; integrated SAM-IS ISP auto-exposure into the Samsung camera framework.
• Debugged link and timing problems across the whole capture chain with Saleae, Introspect and Rohde & Schwarz: C-PHY/D-PHY behavior, pixel and tunnel mode, packet and pipe mapping, frame timing. Caught a wrong deserializer variant on prototype hardware early and kept development running on a 3 Gbps workaround until corrected boards arrived.

Platform integration, tooling & lab

• Brought up camera drivers on NVIDIA Tegra (Jetson AGX Xavier, Connect Tech carrier boards) and Samsung Exynos Auto, including device tree and kernel media-pipeline (Tegra VI / V4L2) integration. Built Solectrix frame-grabber init sequences, a frame-grabber NvM/OTP programming tool (also used to support production programming), a GMSL SerDes link monitoring tool, Image Quality lab tooling and Exynos/NVIDIA build and deployment scripts; contributed to RFQ handling.
• Set up a remote camera-driver development lab (remote power and lighting, IR-LED presence and exposure checks, remote oscilloscope, logic analyzer and flashing), which shortened validation and let the team work remotely on shared hardware. Wrote the automated load/unload stress harness and log analyzers used to qualify driver releases, and a self-check script that spots camera wiring, wrong-port and LED-power faults so other teams can get a camera running on their own.

Ownership & cross-team work

• Camera-driver point of contact for the program: support hardware (incl. EMC testing and HW2SW interface reviews), image quality, validation, homologation, feature and platform teams across several sites, and work directly with external vendors (Samsung, Analog Devices/ST, Solectrix).
• Turn incoming field and integration reports into reproduced, diagnosed and documented issues; review the team's driver code; represent camera software in program status meetings. Team lead of the three-person camera software team for a period.
• Brought GitHub Copilot into the camera team and help colleagues get going with coding agents; HARMAN "AI buddy".

Consultant / Expert Software Engineer · Jun 2019 – May 2022

Mobica (contracted to a leading microprocessor manufacturer), Łódź

• Developed the Linux kernel and userspace driver (x86-64) for a PCI hardware cryptography and compression accelerator, within a 7-engineer team, and was its team lead for part of the engagement; handled PCI device probing, hardware detection and firmware loading from configuration.
• Migrated the driver from kernelspace UIO to userspace VFIO, and implemented PF-to-VF (SR-IOV) communication, sysfs configuration/status interfaces, kernel interrupt handling and PCI Advanced Error Reporting (AER). Diagnosed and fixed timing-dependent race conditions on new hardware revisions, segfaults across the driver and its Software Abstraction Layer middleware, and nightly-test failures.
• Profiled and optimized middleware startup (memory-pool initialization); built kernel-driver unit tests with Google Test/gMock, ran Bullseye coverage and Coverity/Klocwork static analysis, fixed Autotools build scripts and prepared a Fedora/RPM release. Earlier, delivered a benchmarked DPDK-based Vector Packet Processing (VPP) plugin in Linux userspace.

Senior Software Engineer · Jun 2016 – May 2019
Software Engineer · Aug 2014 – May 2016

HARMAN International (incl. Symphony Teleca) – Vehicle CPU Team, Łódź

NTG6 / NTG6@223 / NTG7, Daimler infotainment (Mercedes-Benz MBUX).

• Brought up custom automotive head-unit boards on Renesas RH850 and NVIDIA Tegra: low-level driver development, hardware and software integration, and MCAL configuration.
• Developed embedded C system software and complex device drivers on a real-time automotive stack (AUTOSAR): SPI, CAN / CAN-FD, watchdog, NvM and peripheral drivers (ADC, GPT, ICU, DIO, PORT, ETH).
• Implemented SPI inter-processor communication (shared-memory protocol), time-windowed watchdog task monitoring, and CAN-FD integration for a Continental rear-view camera.
• Before the Symphony Teleca acquisition (2014–2015): developed Android/Linux camera manufacturing tests on the Qualcomm camera framework (MSM8974/MSM8992, Snapdragon) in C/C++ for a phone manufacturer.

Embedded Software & Hardware Engineer · Jul 2010 – Jun 2014

Alitec, Łódź

• Developed embedded software in C (GCC) for ARM, Atmel AVR and Linux, and designed industrial control and measurement hardware (circuits and PCBs) in Altium Designer.
• Built a medical-transport tracking device (TI OMAP/BeagleBoard, Buildroot Linux, AVR watchdog), an RS-485/Modbus strain, temperature and humidity monitoring module (built for an industrial-hall roof, later reused with a Modbus display unit for crane-load measurement), and firmware for a 16-channel 24-bit measurement card (ARM7, Ethernet).

Embedded Systems Engineer / Researcher · Dec 2011 – Apr 2014

Lodz University of Technology – Dept. of Microelectronics & Computer Science, Łódź

• Designed and built the wireless sensor nodes of a distributed structural-health-monitoring and vehicle-weight-measurement system (TULCOEMPA): communication architecture, node hardware (strain-gauge front end, 24-bit ADC, low-power design) and node firmware (Contiki OS on MSP430 for IEEE 802.15.4 mesh routing and power management, ARM Cortex-M4 measurement firmware), plus the OMAP/Linux data sink (Buildroot).
• Presented at international conferences (DGM Best Paper Award), completed a research internship at Empa (ETH Domain), Zürich, and taught undergraduate courses (ARM microcontroller programming, PCB design, digital electronics, Java/C#).

Education

Doctoral (PhD) studies, Computer Science · 2011 – 2017

Lodz University of Technology (Politechnika Łódzka)

Research on wireless sensor networks and embedded Linux for measurement systems, published as first-author journal and conference papers; dissertation not submitted. University scholarships for top doctoral students in three consecutive years (2013–2015) and the Łódź Voivodeship doctoral scholarship "Nauka drogą do biznesu".

M.Sc. Eng., Electronics & Telecommunication · 2006 – 2011

Lodz University of Technology – specialization: Microprocessor Systems & Programmable Devices

Projects, publications & training

Open source

ov5647-jetson: tegracam V4L2 driver and device tree overlays bringing the OmniVision OV5647 (Raspberry Pi Camera v1) to NVIDIA Jetson boards across L4T and JetPack releases, with raw V4L2 and hardware-ISP capture paths. BC-250: GPU compute on the AMD BC-250 (RDNA2 APU, GFX1013, 16 GB unified GDDR6) for local LLM inference (llama.cpp, Ollama, Vulkan) and stable-diffusion.cpp, with a companion bc250-rocm repository on ROCm/HIP and amdgpu kernel behavior.

Publications & talks

Eight peer-reviewed papers (2012–2015, five as first author) on wireless sensor networks and structural health monitoring, in IEEE MIXDES, the International Journal of Distributed Sensor Networks, the International Journal of Electronics and Telecommunications and Microelectronics Reliability (full publication list on Google Scholar). Two articles in Paged Out! Issue #9 (Jul 2026, ISSN 3043-0534): "Your Car Is Watching You: How Camera Pixels Cross Meters of Cable Into Linux" (p. 21) and "Edge AI on old mining silicon: LLMs and Diffusion on 16 GB AMD APU" (p. 35). Co-founder (2017) and organizer of CEHUG Łódź, the local C/C++, embedded and hardware user group; talk there, June 2026: "Your Car Is Watching You: In-Cabin Cameras from Sensor to V4L2 Driver" (talk recording on YouTube, talk slides (PDF)).

Training

GMSL University, Analog Devices (2026); HARMAN AX AI Acceleration Program; Linux Kernel Debugging (Sages); AUTOSAR 4 in Practice (Vector Academy); Altium Designer.

Languages

Polish (native), English (fluent), German (basic).

akandr.cc · pdf · updated September 2026