Philip Shawn OyugaEmail Philip

Philip Shawn Oyuga

One engineer, three layers: firmware, security and machine learning.

Features

  • Firmware written for STM32 and Raspberry Pi Pico
  • Pentested a production fintech product
  • Runner-up in two capture-the-flag competitions
  • Filing a patent on a multi-agent data-extraction architecture
  • BSc Electronics and Computer Engineering, JKUAT, 2025
Fig. 1Block diagram. Drag to turn it, or pick a layer. Scroll to take it apart.

Selected projects

Multi-agent extraction, autonomous robotics, construction software and greenhouse monitoring.

Multi-agent extraction

A multi-agent workflow that extracts healthcare facility records from state licensing websites for Seniorpark.

Seniorpark public homepage with its senior care facility search.
Live siteSeniorpark public websiteseniorpark.ioVisit site (opens in a new tab)
Multi-agent extraction: characteristics
ParameterValue
DatesSep 2025 – now
LayerMachine learning
Tested byAutomation tests on the scraper
Monitored byOpenTelemetry
StatusPatent filing under way

Autonomous 4WD car

A small four-wheel-drive robot that avoids obstacles and finds its target.

Illustration of the autonomous 4WD robot car
Fig. 3
  1. Raspberry Pi Pico controller
  2. Obstacle sensing
  3. Four-wheel drive
Autonomous 4WD car: characteristics
ParameterValue
DatesFeb – Apr 2024
LayerHardware
ControllerRaspberry Pi Pico
DriveFour-wheel drive
DoesAvoids obstacles and locates its target

Atomio

A software company I co-founded. We built construction-management software and client websites, including Laxmi Group and SafeHarbor Consulting.

Laxmi Group website About page, with team, construction and community project photographs.
Laxmi GroupClient sitelaxmigroup.co.keVisit site (opens in a new tab)
Atomio: characteristics
ParameterValue
Dates2024 – Jan 2026
RoleCo-founder
BackendC#, on servers I configured
Client websitesLaxmi Group and SafeHarbor Consulting

Smart greenhouse

A final-year team project combining greenhouse sensor monitoring with tomato-leaf disease identification.

I built the IoT network and contributed to parts of the ML pipeline.

Read the project summary: Smart greenhouse
Tomato plants inside the timber-framed greenhouse, with translucent walls and overhead wiring.A wired electronics assembly with a display labelled temperature and humidity.Young plants in a soil bed beside black pipework and fittings along the greenhouse wall.A hand holding a cluster of green tomatoes still attached to the plant.
Build photos
Smart greenhouse: characteristics
ParameterValue
DatesJan – May 2025
LayersHardwareMachine learning
HardwareSTM32 Discovery and ESP32-CAM, connected over UART
ML pipelineSAM leaf segmentation and TensorFlow CNN disease classification
InferenceCloud-hosted, served through Flask
ContextFinal-year team project

About me

Hi, I’m Philip, also known as shawn-phy. I’m a computer engineer working in machine learning and cybersecurity.

I can’t think of an electronic device I’ve owned that I haven’t opened up to peek inside. By the time I was nine, an engineering degree already seemed inevitable.

I like working across hardware, security and systems development because I like starting from first principles: understanding what’s happening underneath the abstractions, then building from there. That habit has saved me plenty of debugging hours.

I’m always looking for stubborn problems, whether in competitions or hackathons. Some attempts have admittedly been flops; others have become projects I’m proud of.

Off the keyboard, I enjoy playing chess, cycling and fixing electronics.

Hardware and low-level systems

Firmware, boards and the low-level code underneath everything else.

The hardware character

Built with

  • C
  • C++
  • STM32
  • Raspberry Pi Pico
  • Mbed OS
  • PCB design
  • JUCE
Hardware and low-level systems: dated work, most recent first
WhenWhat I built
Jan – May 2025

Smart greenhouse

View project details: Smart greenhouse

Also machine learning

Jun – Aug 2024

RoboDojo competition robot

A robot that finds its way through a maze and collects objects, built in a team of four.

Feb – Apr 2024

Autonomous 4WD car

View project details: Autonomous 4WD car

Sep – Dec 2023

Smart home system

A custom PCB and its firmware, in a team of four.

Aug – Dec 2023

Docker from scratch in C

Containers built from Linux pipes, child processes and syscalls.

Jul – Oct 2023

I2C driver for the STM32F429ZI

Firmware for the VL53L0X time-of-flight sensor, running on Mbed OS.

Apr 2022 – now

Audio filter plugin

A filter plugin for FL Studio, built with the JUCE framework. In progress.

Dec 2021 – Jan 2022

Media server from a 24-year-old PC

Legacy hardware put back to work as an Emby media server.

Security

Breaking systems on purpose: pentests, exploits and capture-the-flag.

The security character

Built with

  • Rust
  • Bash
  • Networking
  • Pentesting
  • Exploit development
  1. Sep 2025 – now

    Fintech pentest

    Pentesting a production fintech product as part of my current role.

  2. May 2024 – now

    Notepad exploit in Rust

    An exploit written in Rust. In progress.

  3. 2022

    International Rungu CTF

    First runner-up!

  4. 2022

    National Cyberise CTF

    First runner-up!

Machine learning

Models and agents that do real work, from a Jetson Nano to multi-agent pipelines.

Built with

  • Python
  • TensorFlow
  • OpenCV
  • Jetson Nano
  • OpenTelemetry
  • Quantum circuits
The machine learning character
  1. Synthetic datasets with qGANs

    Classical autoencoders combined with variational quantum circuits to generate images.

    Jun – Aug 2026

  2. Multi-agent extraction

    View project details: Multi-agent extraction

    Sep 2025 – now

  3. Monitoring for AI workloads

    One OpenTelemetry setup that watches every AI workload.

    Sep 2025 – now

  4. Tennis match live analysis

    Real-time match analysis with OpenCV and TensorFlow, deployed on a Jetson Nano.

    Mar 2024 – Apr 2025Also hardware

  5. SeAn

    Sentiment analysis for WhatsApp chat exports.

    Jun – Aug 2022

Pin configuration

The languages and platforms behind the work on this page.

Close to the metal

  • C
  • C++
  • Rust
  • Bash
  • STM32
  • Raspberry Pi Pico
  • Jetson Nano
  • Mbed OS

Up the stack

  • Python
  • TypeScript
  • C#
  • TensorFlow
  • OpenCV
  • OpenTelemetry
  • GraphQL
  • React
Fig. 6Sixteen pins, all in use.

Experience

Engineering roles, most recent first.

Revision history: roles, newest first
RevDatesRoleWhat changed
DSep 2025 – nowML and cybersecurity engineer, NairobiMulti-agent extraction, a fintech pentest and monitoring for AI workloads.
C2024 – Jan 2026Co-founder, Atomio TechnologiesConstruction-management software, client websites and a C# backend.
BAug 2024 – Apr 2025Junior engineerMoved a PHP monolith to TypeScript microservices and wrapped RouterOS for payments.
AMay – Jul 2024Software developer internA multithreaded financial-statement parser and a C# event system with a GraphQL API.
Awards
  • 2ndInternational Rungu CTF, 2022
  • 2ndNational Cyberise CTF, 2023
  • 3rdNational Science Fair, regional, 2018
Education
  • BSc Electronics and Computer Engineering, JKUAT, 2020 – 2025
Certificates
  • Quantum Machine Learning, advanced level, IBM, 2025
  • International Mathematics Olympiad, participation, 2018
Off the clock
  • Chess
  • CTF competitions
  • DIY builds
  • Cycling
  • Music production

Hiring for work that crosses the stack?

Email me about a role or a project. I am based in Juja, Kenya, and work with teams anywhere.