Core Technical Skills
- Domains: Safety-critical automotive (Classic AUTOSAR, ISO 26262, ASPICE), embedded systems, applied machine learning
- Programming Languages: C/C++, Python, TypeScript, AssemblyScript, Bash
- Embedded & Quality: Infineon AURIX, JTAG, hardware-in-the-loop testing, ASAN/TSAN, Valgrind, Coverity, SonarQube
- Build & CI: Bazel, CMake, Docker, GitHub Actions, Zuul
- Languages: Portuguese (Native), English (C2), Spanish (B1), German (A2, studying for B1)
Work Experience
Software Engineer – Critical Techworks
January 2024 – Present
Crowd Data Collector is BMW’s project for in-vehicle data collection. Its most unusual component is an “app runtime for the car”: teams deploy small sandboxed workloads over the air onto vehicle ECUs. New features do not have to wait for a firmware release. I have worked on both sides of it, the runtime itself and the jobs that run on top of it, together with teams in Germany and China. Alongside that, I designed and ran an internal coding competition for the team, written up on the hackathons page.
Classic AUTOSAR Job Runtime
January 2026 – Present
The job runtime on BMW’s Heart of Joy, a Classic AUTOSAR ECU built around an Infineon AURIX microcontroller. The same ECU runs the driving dynamics and manages the high-voltage battery, so the runtime must never disturb the applications around it. My team took it over from the team in China that built it.
- Observability: Designed and integrated a per-job metrics service, the first new feature on the runtime after the takeover.
- On-Target Validation: Assembled the hardware test bench and validated the complete path on the ECU with hardware-in-the-loop tests and a JTAG debugger. Improved the end-to-end test framework.
- Tooling: Added persistent storage to the host-side simulator, so stateful jobs can be tested across restarts without hardware. Maintained the release and CI pipelines across Linux, macOS and Windows.
- Reliability: Integrated lifecycle and concurrency fixes into a vehicle release and fixed defects found during integration.
Edge Computing & WebAssembly Platform
September 2025 – Present
A hardware-agnostic C++14 and POSIX framework that orchestrates sandboxed WebAssembly workloads across several ECU architectures. It is deployed on every BMW Neue Klasse vehicle.
- Sandboxed Execution: Built permission-gated system services in the WebAssembly runtime. A job is compiled once and then runs sandboxed on every supported ECU.
- Job Lifecycle: Contributed to the cross-ECU job lifecycle and its API, covering install, validation, update and uninstall.
- Production Hardening: Hardened the runtime with ASAN, TSAN, Valgrind, Coverity and SonarQube, run through the Zuul CI pipelines. Added rate limiting and RAM monitoring to keep jobs inside the memory budget of constrained hardware.
- Observability: Added metrics to the rate limiter, so the team can see which jobs hit their limits, and surfaced each job’s most recent errors so the backend links directly to the relevant logs. Built event-loss and received-event metrics in the platform’s event pipeline.
- Testing: Wrote concurrency stress tests for a platform service and fixed the deadlock they exposed under ThreadSanitizer. Removed coverage exclusions across the codebase and covered the gaps with unit tests, so the coverage gate covers all code.
- Release Automation: Automated releases for a shared library.
In-Vehicle Data Collection Jobs & SDK Ecosystem
March 2024 – September 2025
- Data Products: Designed and shipped over a dozen in-vehicle data-collection jobs across BMW’s electric fleet, several of them end to end: trigger and activation logic, signal processing, and delivery to the backend.
- SDK & Code Generation: Co-designed and built most of a TypeScript SDK that turns CAN and FlexRay signal definitions into typed AssemblyScript code. Jobs receive named, typed signals instead of decoding raw payloads by hand.
- API Evolution: Kept the generated code in step with the platform APIs as they changed across vehicle programmes and releases.
C++ Academy
January 2024 – March 2024
Three-month intensive C++ programme at the start of the role, covering OOP, concurrency, memory management, and CI/CD with Docker and GitHub Actions.
Academic Experience
Research Scholarship – GreenAuto Programme – DIGI2 Laboratory, FEUP
December 2022 – October 2023
My M.Sc. dissertation, done on a research scholarship within the GreenAuto project for automotive industry sustainability. Designed a low-cost wireless sensor module for fault detection in Automated Guided Vehicles and nearby machinery. The full dissertation and presentation are on the project page.
- Embedded Firmware: Developed Teensy firmware to sample vibration, sound and temperature at fixed rates, with SD card logging, and tested WiFi transfer to an InfluxDB database.
- Fault Detection: Built three lab testbeds and compared eight machine learning models. XGBoost detected a poorly lubricated bearing with an F1 score of 0.99, and reached 0.74 on a testbed outside its training data.
- Data Processing: Selected compact statistical features that could later run on the sensor module itself, reducing wireless traffic.
Summer Internship – ML Toolkit – DIGI2 Laboratory, FEUP
July 2022 – September 2022
A Python toolkit that finds the input parameters producing a desired target output for a regression problem.
- XGBoost regressors with hyperparameter tuning through Hyperopt, and simulated annealing (SciPy’s dual_annealing) to search the parameter space.
- SHAP feature-importance analysis, so the model could be inspected rather than taken on trust.
- A Streamlit interface, with the machine learning kept separate from the UI so that it could be driven from another frontend.
Volunteering
President of the General Assembly – BEST Porto
2022 – 2023
Chaired the general assembly of BEST Porto, the local branch of the Board of European Students of Technology.
EBEC Porto 2022 – Challenge Lead
October 2021 – May 2022
Designed and ran the Team Design challenge for EBEC Porto 2022, the largest engineering competition in Portugal, in partnership with Saltpay.
- Wrote the challenge itself: teams built an ATM prototype handling withdrawal selection, card insertion and token dispensing.
- Coordinated more than 200 participants over the 24-hour format at FEUP, and provided the on-site electronics support, from circuit troubleshooting to failing hardware.
Achievements
- Top 10 – Hackacity 2023: Smart city data challenge on CO₂ reduction (Porto Digital)
- 2nd Place – Datattack 2023: 24h data science challenge for Civil Protection (IEEE Student Branch Porto)
- 3rd Place – EESTEC Challenge Porto 2022: Machine learning competition on colour blindness detection
Education
- M.Sc. in Electrical and Computer Engineering – FEUP (2021–2023)
- Specialisation: Industrial Automation, Embedded Systems, and Robotics, including IEC 61131-3 PLC programming
- Dissertation: Sensor System for Predictive Maintenance in Industrial Environments
- B.Sc. in Electrical and Computer Engineering – FEUP (2018–2021)
- Erasmus+ Exchange Semester – Universidad de Sevilla (2022–2023, during the M.Sc.)