Llamas, Machine Learning and a Trip to Kenya
Sharing Voyage’s learnings with new communities


At Voyage, communities are at the heart of what we do. Our autonomous taxi service provides safe, accessible transportation to our amazing partner communities — and we learn something new each time a passenger gets in the car.
Fostering relationships with new communities — and giving back a little bit of what we’ve learned along the way — is a big part of what makes our team at Voyage excited to come to work.
Recently, I had the opportunity to co-organize and teach at the Data Science Africa (DSA) summer school and workshop in Nyeri, Kenya. This was my second trip to DSA, and I was excited to contribute some of my experiences from Voyage to this year’s “end-to-end data science” curriculum.


Data Science
What is “end-to-end” data science? It means rethinking traditional data science workflows and embracing new principles that deeply involve the people and communities being studied. This includes best practices like:
- Involving community members in the selection of study questions
- Partnering with communities in the data generation and curation process
- Being intentional when formulating models, assumptions — can we think like engineers rather than scientists?
- Rigorous, controlled experimentation
- Transparent communication of results back to the community and stakeholders (e.g. policy makers)
At DSA, our aim is to have every participant, student, and teacher embody these aspects in their practice of data science. We do this by using practical datasets that attendees can identify with, involving domain experts during the data science process, and always seeking to incorporate the newest best practices.


Data Science Africa
What’s it like to attend DSA? In a word: awesome.
DSA Nyeri was chaired by the excellent Ciira Maina and Earnest Mwebaze. The event involved three field work activities in addition to a series of deep dives into the latest in machine learning and IoT techniques. Topics covered everything from the very basics of Python, to CNNs, Reinforcement Learning, and real-world model deployment techniques.
One of the best parts of DSA is the broad, diverse group of attendees. Participants came from all over the African continent and far beyond. The students fell on a large spectrum of professional experience, ranging from undergraduate students to senior practitioners and lecturers. The event was held at the Dedan Kimathi Univerisity of Technology in Nyeri, Kenya, aptly named after one of the most prominent freedom fighters in Kenyan history.

DSA was taught by a group of world-renown experts. Field work included a greenhouse sensor deployment, a camera-based animal classification ‘trap’, and air quality monitoring. Thanks to ARM, we had access to custom, DSA-branded Mbed boards for these deployments, thanks to Jan Jongboom. The deployment of air quality sensors meant we got visit to the university’s amazing wildlife conservancy — during which we met llamas!


Because we had a computer vision session lined up the next day, we could not resist the temptation of testing classification on the llamas. We downloaded a pre-trained YOLOv3 model and immediately tried it out. Unfortunately, we were disappointed. It predicted a giraffe!
The “unknown llama” experience inadvertently became a great discussion point for DSA attendees. It reminded us of the need for inclusive, relevant datasets — of course our local community members knew about llamas! It also highlighted an industry-wide need for “data sheets” for ML models (like those for datasets), as we had to experiment before learning the model had probably never seen a llama.
An amazing trip to the wildlife conservancy left us with some great lessons for the students!
DSA Nyeri was an excellent event. In the span of a few days, we created our own small community, with new friendships, experiments, and collaborations forged. In addition, we benefitted greatly from the support of the local Nyeri community, and sponsors like Voyage who enabled their employees to volunteer and share their knowledge with students.
Interested in getting involved with the DSA? The best way is through participation. Companies can sponsor the event, send employees as volunteers, and provide learning opportunities for students. And if you’re an individual data scientist, you can learn about opportunities here.
The next DSA will be held in Abuja, Nigeria.
Interested in learning more? Additional day-by-day account has been given by the amazing Damon Civin.


Last but certainly not least, I had the privilege of doing a fireside chat on AI and self driving cars at the Nairobi Women in Machine Learning and Data Science (WiMLDS) meetup in Nairobi.


