Guilherme Brandão da Silva

MSc. in Electrical Engineering |

Embedded Software Engineer at Wind River

I am highly passionate about research and development in new technologies, constantly seeking out challenges. I find great excitement in writing software to solve real‑world problems, while not being afraid of any hardware integration. Thrive in diving into complex code bases and solving problems others may avoid. I possess strong communication skills, driven by teamwork, and always willing to assist those around me while objectively reporting results. I am a fast learner with experience in Linux, Android, and bare‑metal embedded systems development. My usual activities include real‑time programming, telemetry, computer vision, and electronic instrumentation.

Education

Experience

Wind River (Embedded Software Engineer | 09/23 - current)

Working with Critical River. Contributed to code hardening efforts to achieve DO-178C compliance for the VxWorks’s BSPs (Board Support Package) and Bootloader. Designing critical software, API specifications, and conducting static analysis to ensure adherence to certification guidelines for safety and quality.

• Code hardened and complied with certification guidelines for safety and quality. Updated, built, debugged, and tested VxWorks components. (C, VxWorks, MISRA, Coverity, Jenkins)
• Worked on the API specification for several components used for DO-178C certification evidence. (C, DO-178C, Polarion)
• Create and upgrade developer tools for static analysis, binary analysis, and code hardening. (Python, DWARF, Elftools, Jenkins)

Golfleet/V3 (Embedded Tech. Lead. | 03/21 - 09/23)

V3 provides TSPs and OEMs market with end‑to‑end video telemetry solutions applied to smart dash‑cams besides API integration for fleet management.

Embedded Technical Leader (03/22 - 09/23)

• Designed and implemented the event‑driven software architecture for the principal product. The new version presented a reduction of almost 80% of the code, and the memory profile was reduced by 61%; at the same time that included new key features for the product. (Android Headless, OOP, Design Pattern, Gherkin)
• Developed testable and flexible software for smart dash‑cams. Worked on the sensor’s data acquisition and processing for analyzing vehicle and driver’s behavior and other modules. (Java, Kotlin, Python, Pytest, Mockk, Android Headless)
• Implemented communication protocols and message generation system. Designed device telemetry events for TSP companies. Also, the over‑the‑air updates, commands, and configurations for the devices. (Java, Kotlin, ProtoBuffer, AWS, UDP, MQTT)
• Designed the Q&A process and tools. Automated CI/CD pipeline, the end‑to‑end, and HiL (hardware‑in‑the‑loop) equipment tests platform. Created tools, scripts, and software integration for analysis and deployment. (Python, Pytest, Gitlab CI/CD) – Acted as technical leader on project management, research on new technologies for the products, organized team activities, and planned milestones. (Agile, Scrum, Gitlab Project). – Conducted meetings with other teams, requirements analysis, commercial presentations, and technical reports. Mentored other engineers. Helped to build the developer’s team, lead interviews, and hiring.

Embedded Software Developer (03/21 - 02/22)

• New features development, technical improvements, and maintenance of existing embedded software.
• Porting of existing embedded applications to a new hardware device.
• Development of services, APIs, and libraries for cloud applications.

Ecozone (Embedded System Engineer | 2019 - 2020)

Electronic system for instrumentation of PH3 and O3 concentration and control of the silos purging process.

• Embedded platform design for real-time data telemetry.
  • Firmware architecture planning for the sensors instrumentation and the real-time data acquisition and telemetry. (C, FreeRTOS, STM32-HAL)
  • I worked on the PCB design for the sensors boards and the embedded platform, which had an MCU (ARM Cortex M), serial communication (such as SPI for the sensor interface, and RS485 to communicate with the control module), and actuators.
• Developed the central board for real-time monitoring and control of the process (HMI).
  • Custom Linux kernel for the embedded system.
  • Developed the data acquisition, and processing for monitoring the purge process. (Embedded Linux, C/C++).
  • Graphical HMI development for assisting the data acquisition (telemetry), visualization, process monitoring, and control. (C++, Qt5).

Lapaza (Embedded System Engineer | 2018 - 2019)

Embedded System design for a Smart Parking machine. The hiring company had similar devices and was interested in updating some of its hardware components and adding new software features.

• Custom Linux kernel development, with different peripherals such as a thermal printer, coin counter, 16x4 LCD display, RFID reader, keyboard, electronic actuators, and communication interfaces (WiFi, GPRS, etc).
• Software application development for smart parking. It involved handling tasks such as system operation modes design (state machine), sensor readings, activation of the actuators, cloud storage, etc.
• Tweaked the embedded OS (Linux) to redirect terminal outputs, disable not used peripherals, etc, to protect the company’s Intellectual Property.
• Assisted in the PCB redesign for the embedded system operation requirements.

Projects

This work proposed the heterogeneous embedded implementation of a real-time lane detection system capable of estimating lane marks from images of a monocular camera. Each frame is processed to obtain a suitable bird’s eye view perspective. Were extracted the lane mark features to build different feature maps. Then the maps are combined to generate a robust binary frame with the lane marks used for the lanes estimation.

• Responsible for the embedded hardware choice, algorithm design, and software implementation, testing, and validation (CUDA, C++, OpenCV, GoogleTest, VisualProfiler, Python).
• The software development and evaluation were on an embedded Jetson Nano board (an ARM A57 CPU and Maxwell 128 CUDA cores GPU). The whole application was implemented in C++/CUDA programming languages achieving high performance in heterogeneous programming. The GoogleTest was used for tests and the TuSimple dataset for the benchmarks and evaluations.
• The developed heterogeneous software performance ranged from 92.3% to 99.0% accuracy in the datasets, detecting up to 97.9% of the ego lanes available. The embedded system achieved real‐time constraints, running 25 and 140 times faster than the hybrid and CPU‐executed, respectively.

The work proposed an end-to-end test environment for modeling and controlling quad-copters based on hardware-in-the-loop simulation. Responsibilities and key achievements:

• Responsible for the embedded hardware, firmware implementation, model simulation, and HiL platform.
• The design and implementation of quad-copter dynamical and kinematic models and control plants for an online simulation. (Matlab/Simulink)
• A graphical interface to control and watch the telematics data from the under-test quad-copter ControlDesk (dSPACE).
• The development of firmware for embedded quad-copters. Capable of real-time sensors telemetry, motors control, and receiving the online configurations and commands from the HiL platform. (C, FreeRTOS, STM32-HAL, Electronics)
• The 4DoF structure provides a safe and realistic test bench for the end-to-end test of quad-copters.