Andrei Cîrcei — Software Engineer

Embedded systems & automation

Andrei
Cîrcei

I build embedded systems and the automation around them — low-level C, Python tooling, and clean delivery pipelines.

Currently @ e-novation GmbH
  • C
  • Python
  • CAN
  • CI/CD
RO · 2024 →

01 About

I like building efficient, scalable systems and solving real-world problems with software that does its job quietly and well.

Open-minded and always learning — I think steady curiosity is what actually drives good engineering. Across a few different roles I've also picked up the practical things: working in a team, staying level under pressure, and managing my own time and priorities.

These days my focus is embedded development and automation: writing low-level C, building Python tools and scripts, analysing vehicle communication, and keeping CI/CD pipelines healthy so integration and delivery stay smooth.

02 Experience

  1. 2024— now

    Software Engineer · e-novation GmbH

    Embedded software and the automation around it, across several automotive programs — commercial-vehicle, passenger-car and lighting. I build hardware-in-the-loop test frameworks and the GitLab CI/CD that runs them, work down at the level of UDS diagnostics, secure flashing and ECU debugging, and up at model-based code generation and in-vehicle communication. Lately I've been wiring AI tooling into the hardware loop. Day to day: low-level C, a lot of Python, and Lauterbach / Vector tools on the bench.

    • C
    • Python
    • UDS / CAN
    • AUTOSAR
    • Lauterbach
    • GitLab CI/CD
  2. 2022— 2024

    Training Instructor, IT · Logiscool Brașov

    Part-time. Taught programming and computational thinking to kids and teens — from visual logic blocks up to JavaScript, C#, and Python — and worked on sharpening their analytical thinking.

    • JavaScript
    • C#
    • Python
    • Teaching
  3. 2020— 2022

    Data Collection & Recruiting · FocusData

    Freelance, part-time work during high school — sourced multilingual voice recordings from contributors worldwide to help train AI speech systems.

    • Data collection
    • Recruiting
    • Remote

03 Selected work

A look at the kind of problems I work on, grouped by theme. Client names are withheld; each sits in automotive — commercial-vehicle, passenger-car or lighting.

  1. Commercial-vehicle ECU program

    Hardware-in-the-loop test automation

    Designed and grew a Raspberry-Pi-based hardware-in-the-loop test framework for automotive ECUs — UDS diagnostics, automated flashing, core-dump validation and release/smoke testing — all driven from GitLab CI/CD in Python, and used as the team's release gate.

    • Python
    • UDS
    • CAN / LIN
    • GitLab CI/CD
    • Raspberry Pi
  2. Commercial-vehicle & lighting programs

    Secure flashing & diagnostics

    Built ECU flashing and secure access from the ground up: UDS programming sessions, certificate-based authentication (PKI), seed-&-key, and CRC / E2E message protection implemented to AUTOSAR and J1850 profiles.

    • UDS
    • PKI / certificates
    • E2E / CRC
    • AUTOSAR
    • Bootloader
  3. Passenger-car & lighting programs

    CI/CD platform engineering

    Built and optimized GitLab pipelines for embedded build, secure-boot signing and measurement-data processing — parallel runs, diagnostic (ODX / PDX) build automation, plus notification and error-aggregation tooling. Cut one data trace from 36 GB to 0.6 GB and trimmed pipeline runtime and resource use.

    • GitLab CI/CD
    • Docker
    • Conan
    • ODX / PDX
    • Data processing
  4. Commercial-vehicle ECU program

    Low-level embedded debugging

    Root-caused deep faults on 32-bit automotive microcontrollers — NMI traps, hardware safety-monitor alarms, and memory-layout / DMA issues — using a Lauterbach debugger, crash-log decoding, linker maps and on-chip breakpoints.

    • AURIX / TriCore
    • Lauterbach Trace32
    • Crash-log analysis
    • Assembly
  5. Commercial-vehicle SW platform

    Model-based software & code generation

    Worked on a Python architecture-generation toolchain: meta-model → ARXML, plus COVESA IFEX / VSS / Protobuf and SOME/IP deployment configs for a service-oriented vehicle platform on QNX. Benchmarked SOME/IP against zero-copy IPC (iceoryx) and uProtocol.

    • AUTOSAR / ARXML
    • COVESA (IFEX / VSS)
    • SOME/IP
    • QNX
    • Python
  6. Commercial-vehicle & passenger-car programs

    AI-assisted engineering tooling

    Built MCP (Model Context Protocol) servers that expose a hardware debugger and a bus-analysis tool to AI agents — cutting per-call token use to tens or hundreds of tokens — and used LLM-driven generation to turn requirements into executable test cases. A related diagnostic-data automation cut a manual task from hours to seconds.

    • MCP
    • LLM tooling
    • Python
    • Automation

04 Education & certifications

05 Stack

Languages

C · C++ · Python · Ruby · CAPL · XSL/XSLT · Bash · SQL

Vehicle & diagnostics

UDS · CAN / CAN-FD · LIN · Automotive Ethernet · SOME/IP · E2E protection · XCP

Embedded & AUTOSAR

AURIX / TriCore · AUTOSAR · ARXML · bootloader & secure flashing · QNX · Linux · Raspberry Pi

Debug & test

Lauterbach Trace32 · Vector CANoe / CANalyzer · HIL test automation · crash-log analysis

Automation & CI/CD

GitLab CI/CD · Git · Docker · Conan · measurement-data pipelines

AI tooling

MCP servers · agent skills · LLM-assisted test generation

Languages spoken

Romanian (native) · English (C1 Advanced)

06 Contact

Let's talk.

Happy to talk embedded systems, automation, or any interesting engineering problem. Email is the best way to reach me.