Richard Djarbeng
Software Engineer | ML Engineer | Robotics & Embedded Systems
I am a software engineer and machine learning engineer with an MSc in Electrical and Computer Engineering from Carnegie Mellon University. I build low-cost sensing systems and machine learning pipelines for environmental monitoring: AI-calibrated sensor networks in East Africa, a smartphone-based method for measuring black carbon, and the web dashboards that make the data usable. I work across computer vision, time-series modeling, and data engineering, and I am currently preparing PhD applications centered on machine learning for environmental sensing and air quality. AWS Certified Solutions Architect. I also write technical posts and personal reflections on this blog.
Experience
Research Associate
Feb 2024 - Oct 2025Worked with Prof Edwin Mugume on a project building low-cost weather sensors, focusing on the deployment and maintenance of these solutions across East Africa.
Technical details
- 3D radiation shield design in Fusion 360 for environmental sensor nodes
- Energy consumption and network traffic estimation and simulation for environmental nodes
- Time-series data pipelines and ingestion for sensor networks deployed in the field
- Real-time monitoring dashboards for visualizing air-quality and weather data
Research Assistant
2023 - 2024Worked with Prof Albert Presto's group through the CMU-Africa student exchange. Contributed to a smartphone-based method for measuring black carbon, analyzing photos of Aethalometer samples across multiple wavelength channels to identify the presence of fossil-fuel from wood-smoke pollution.
Teaching Assistant, Internet of Things
Aug - Dec 2023Ran recitations and mentored students building IoT systems: embedded sensors on Raspberry Pi, Node.js backends, and REST/HTTP calls over 2G/3G modems.
Graduate Intern
May - Aug 2023Designed and built a wireless data-logger that automated air-quality data collection from the Thermo Fisher TEOM system, using Python and C++ on Raspberry Pi and Arduino. Built a real-time dashboard in React.js and TypeScript for visualizing air-quality data from live sensor feeds.
Education
MSc Electrical and Computer Engineering
2022 - 2024Specialized in Machine Learning and Software Engineering. Served as Club President for the Internet of Things (IoT) club.
Selected graduate coursework
- Cognitive Robotics (18-799): developed robot software with ROS in C++ (packages, topics, publishers, subscribers), wrote CRAM cognitive-architecture plans in Lisp, and programmed pick-and-place tasks for the PR2 and Lynxmotion AL5D manipulators in Gazebo and Bullet simulators
- Computer Vision (16-720): image filtering and edge detection, HOG descriptors, Lucas-Kanade optical flow tracking, homographies and image warping, camera models, two-view geometry, structure from motion, CNNs, and photometric stereo
- Autonomous Driving (18-744): sensing, perception, localization, driving behaviors, planning, control, and fault management across a full autonomous vehicle stack
- Introduction to Deep Learning (11-785): RNNs/LSTMs, sequence-to-sequence, attention and Transformers, vision architectures (e.g., Vision Transformers)
- Internet of Things: wireless embedded sensing systems and communication protocols
BSc Computer Engineering
2017 - 2021Gained a solid foundation in computer engineering principles and practices.
Electives
- Computer Graphics, with hands-on projects in Unity3D
- Software Engineering, with projects in PHP, JavaScript, Java, and JavaFX across Postgres, OracleDB, and SQLite
- Computer Aided Design, with 3D modeling in Solid Edge (Siemens)
Ongoing Learning
Deep Learning & Computer Vision
Independent study focused on deep learning and computer vision, building models and writing up the process on the blog as I go.
Technical areas I've been studying
- CNNs and transfer learning, using PyTorch and FastAI
- Object detection, semantic segmentation, and video understanding
- Supervised learning and classification, plus time-series analysis
- Reinforcement learning
- Computational linear algebra
Projects
Featured works in Software Engineering and Research
\* Note: Some images are illustrative representations and may not be direct screenshots of the project.
Research Trend
Air Quality Analysis Dashboard
Developed an interactive dashboard to analyze PM2.5 air quality trends in Kigali, Rwanda, using data from US Embassy reference monitors. Found 57% of readings above the WHO "Unhealthy" threshold, with an average of 43.31 ug/m3.
Web App
Countdown PWA
A feature-rich Countdown timer Progressive Web App (PWA) with customizable themes and offline support. Achieved significant user engagement through user-centered design iterations.
Machine Learning
Mapping Filter Imagery to PM Values
Built models to map image data of particulate matter filters, captured across wavelengths, to PM values, providing a low-cost alternative to laboratory analysis for black carbon estimation.
GCP Dataflow
Real-Time Taxi Fleet Management
Designed a streaming data pipeline using Google Cloud Dataflow and BigQuery to ingest and analyze real-time taxi data, displaying live metrics on Looker Studio.
Deep Learning
Free Background Remover
Deployed a semantic segmentation model on Hugging Face to remove image backgrounds automatically. Uses `rembg` and U2Net models for high-quality segmentation.
AutoML
Predicting Loan Risk
Built and deployed a loan default risk prediction model on Google Cloud Vertex AI AutoML. I evaluated it with precision/recall and ROC curves, and checked feature importance to understand the contribution of each predictor .
Big Data
E-commerce Purchase Prediction
Built a classification model on Google Analytics data to predict returning-customer purchase intent. Feature engineering around checkout progress, traffic source, and device moved ROC AUC from 0.72 to 0.94.
Embedded C++
GSM HTTP Requests
Implementation of HTTP GET and POST requests using GSM modules with C++. A core component for remote IoT data transmission.
Java/TSQL
University Admission System
A comprehensive Admission Management System built with JavaFx and TSQL, designed to streamline the university enrollment process.
Remote Sensing
Transmission Corridor Monitoring
Used Google Earth Engine and CNNs to detect encroachment on power transmission corridors in Ghana, enhancing safety for GRIDCo.
Unity / C#
Escape the Covid Maze
A 2D Unity3D game challenging players to navigate a maze while avoiding dynamic enemies using A\* search algorithms.
Open Source
Machine Learning Complete
Contributor to a comprehensive machine learning educational package to democratize AI education.
Notable Open-Source Contributions
Selected contributions to open-source projects
Machine Learning Complete
PR merged September 2023Merged pull request #9: added a curated collection of freely accessible Python learning resources (books, online courses, and videos) to the introductory chapter, verifying each resource was genuinely free before inclusion.
SveltiaCMS
ContributorReported and followed through on several closed issues during development. This website's content workflow also runs on SveltiaCMS.
Areas of interest:
Education:
- What are the best ways to learn AI and machine learning, and what resources are recommended?
- Who are the key people in the field of AI and machine learning, beyond the media hype?
I cover this in posts like Complete Machine Learning Package — Learn by Doing and Machine Learning Specialization by Andrew Ng and DeepLearning.AI:
Complete Machine Learning package- learn by doing
Are you fascinated by machine learning but have no idea where to start? Do you find yourself jumping between YouTube tutorials, blog posts, and dense textbooks, only to end up...
Machine Learning Specialization by Andrew Ng and DeepLearning.AI
The Machine Learning Specialization, created by Andrew Ng and DeepLearning.AI, is a revamped and modernized version of the original Stanford Machine Learning course, one of the most popular and foundational...
Machine Learning Experts to Follow
Looking for inspiration in the field of Machine Learning? Here are my top picks for experts whose work is shaping the future of AI. I’m a huge fan of Andrew...
AI / Machine Learning:
- What are the latest advances in AI and machine learning and what are the best ways to apply them to real-world problems?
- How do you leverage the latest advances in AI using available tools such as cloud (AWS, google) and companies such as Nvidia, Microsoft, Anthropic, OpenAI and AMD?
- Computer vision: What are the latest advances in computer vision and in what areas can they be applied?
Remote Sensing & GIS:
- The Measurement Gap: How do we bridge the “spatial gap” between high-accuracy ground stations (point measurements) and global satellite data (area-averaged)?
- Sensor Fusion: How can low-cost IoT sensor networks (which are often “noisy”) be calibrated/updated in real-time using data from high-grade reference instruments?
- How do ground air quality measurements compare with remote/satellite derived measurements and how can they be improved especially in Africa?
- How do the latest advances in computer vision advance remote sensing and what areas can they be applied to improve sensing?
- What is the impact of advances in Internet of Things (IoT), AI and remote sensing and which initiatives are taking advantage of them especially on the african continent?
Finance:
Why can I pay for Netflix instantly, but not send money to a friend abroad? Why are international payments slow, expensive, and opaque? I explored this phenomenon and potential solutions in this series:
Why Can't I Just Send Money With My Visa Card?
Banks often promote Visa and Mastercard as efficient tools for swift, seamless transactions. However, one critical feature is notably absent: the ability to transfer funds directly between cardholders. This post...
Why Can I Pay for Netflix Instantly, But Not Send $5 To My Brother Abroad? Enter Revolut
In my last post, I wrote about a weird paradox: paying for Netflix with a card is a seamless, one-second breeze, but sending $100 to a friend or family member...
What is Google Pay and Apple Pay? How They Work, Where They Work, & Why They're Safer
In a world where your phone is your wallet, Google Pay (also known as the Google Wallet) and Apple Pay have become essential tools for millions. But what exactly are...
Technical Skills
I work in Python, JavaScript/TypeScript, and C++. I use PyTorch, Keras, and FastAI for machine learning, React for the web, and I deploy models to Vertex AI, BigQuery ML, AWS, and Hugging Face Spaces.
In more depth
- Classification and supervised learning: from logistic regression to CNNs (ResNet) and transfer learning; ROC/PR evaluation; feature engineering.
- Computer vision: object detection (YOLO-World), semantic segmentation (U-Net, ISNet), video analysis, satellite imagery in Google Earth Engine.
- Real-time pipelines and dashboards: streaming ingestion through Dataflow into BigQuery, live dashboards in Looker Studio.
JavaScript (ES6+)
TypeScript
Python
Java
C++
C
C#
Ruby
Dart (Flutter)
PHP
SQL
React.js
GatsbyJS
HTML5
CSS3
PWAs
Performance Optimisation
Responsive Design
Node.js
PHP
Express.js
RESTful APIs
WebSockets
Ruby
AWS (Solutions Architect)
Google Cloud Platform
BigQuery
Dataflow
Netlify
Render
Hugging Face Spaces
CI/CD
Git
GitHub
Open-Source Maintenance
PR Review
Test-Driven Development (TDD)
Agile Methodologies
Certifications
Professional credentials and verified achievements
Download Resumes
Connect With Me
Feel free to reach out for collaborations or just a chat about technology!
Send a Message
Have a question or want to connect? Please add a way to contact you.