The Capacity Accelerator Network is building a workforce of purpose-driven data practitioners worldwide and enabling social impact organizations to unlock the power of data to meet their missions. Forune Adekogbe is a Lead Data Scientist at the University of Lagos where he is working on a machine translation project aimed at enhancing internet inclusivity by providing a way to translate available online information into Nigerian Pidgin.
When did you first realize the potential for using your data skills to make significant impacts in areas like financial inclusion, sustainability, and food security?
Midway through my undergraduate study, I joined a community at the University of Lagos known as the Nigerian Society of Chemical Engineers (NSChE) hub. This hub aimed to help students apply their chemical engineering skills to solve problems related to the Sustainable Development Goals (SDGs). That cohort focused on leveraging waste by converting trash into cash, and my team designed a process that converted waste plastics into 3D printing filaments.
During this project, I realized that sorting waste was a significant challenge, typically requiring substantial human involvement for effective management. Recognizing that this could also be approached as a data problem, I began looking into ways to use my data skills to solve the waste sorting problem. This was a computer vision problem. I realized that automating the sorting process would speed up everything, allowing us to process more waste and contribute to a circular economy by creating value. This marked the evolution of my interest in social impact from a process engineering-focused approach to one where I also applied my data skills to solve societal challenges.
Because of living standards, many skilled workers are motivated by survival and comfort rather than social impact. Therefore, more must be done to recruit, train, and retain data practitioners in social impact.
Fortune Adekogbe Lead Data Scientist University of Lagos
Could you describe a specific project at the University of Lagos NITDA Hub that focuses on data for social impact?
One interesting project from the University of Lagos National Information Technology Development Agency (NITDA) hub that focused on data for social impact was a machine translation project aimed at developing a two-way English-to-Nigerian Pidgin translator. The goal was to enhance internet inclusivity by providing a way to translate available online information into Nigerian Pidgin, a widely spoken language in Nigeria. This would make the internet more accessible to people from various backgrounds who do not typically speak English, allowing for communication in Pidgin. We had to achieve this with minimal resources so that it would be easier for the public to use it across various platforms. We also had to collect data because the data used for the task by previous researchers was no longer available.
In the end, the model we built at the NITDA Hub outperformed existing systems, and we made it publicly available. The diverse data we collected made the model adaptable to different contexts, and we also made it openly available on the NITDA Hub Huggingface page. We presented this project at both national and international conferences and received overwhelming positive feedback, which was very encouraging. It has also been interesting to see individuals and groups use our models for a variety of purposes.
What are some of the challenges specific to data and AI for social impact work?
The availability of data is, unsurprisingly, a major challenge when conducting social impact work with data and AI. Although we live in the internet age, where it appears that you can find anything with a simple Google search, gathering data for specific projects is often difficult. Without available data, progress stalls—we can only theorize. As a result, a lot of effort is put into identifying a data source and developing systems to transfer that data into a warehouse or database. We at the NITDA Hub expect this to some extent based on our experience, but the difficulty may surprise us.
Another challenge is the availability of computing resources. Using AI technologies like machine learning to solve data-related problems necessitates significant computational resources, which are not always available. This limitation makes it difficult to create models that achieve performance goals. Data professionals anticipate problems with processing power, such as GPUs, CPUs, and storage. Still, the scope of the problem is often revealed only after we define our methodology and begin implementing it.
Finally, putting together the right team can be challenging. Because of living standards, many skilled workers are motivated by survival and comfort rather than social impact. This limits their ability to commit to long-term projects, especially if they are volunteers rather than paid employees. While we anticipate this, particularly at the NITDA Hub, and optimize the selection process, life happens and people’s priorities change. Therefore, more must be done to recruit, train, and retain data practitioners in social impact.
The ability to generalize by applying concepts learned in one field to another has been a critical skill I have had to develop to work effectively across multiple sectors.
Fortune Adekogbe Lead Data Scientist University of Lagos
With NITDA, you’ve collaborated across academia, social impact, and non-government organizations to advance data and AI innovation. What key skills have you developed to effectively work across sectors?
The ability to generalize by applying concepts learned in one field to another has been a critical skill I have had to develop to work effectively across multiple sectors. This skill enables me to build on my existing knowledge and combine it with sector-specific insights to help my team at the NITDA Hub achieve its objectives.
Another important skill has been communication, which includes public speaking. Effective problem-solving across sectors requires interaction with others, which necessitates strong communication skills to convey my perspectives and the nature of my work, as well as ensure clarity in my explanations, especially to those outside my domain. Most of the people I work with are not data scientists, so I must communicate complex technological concepts in a way that is engaging and straightforward to maintain their interest and understanding. This skill is also useful in public speaking, where understanding the audience’s level of knowledge is essential for making my presentations accessible and impactful.
Leadership is another important skill that is closely related to communication. I often find myself leading teams at the NITDA Hub, and as a leader, it’s vital to communicate the project’s vision, inspire team members, and foster an environment where they can express themselves freely. Effective team management ensures that our objectives are met on time and that no one feels overworked or dissatisfied. These skills have come in handy as I work with the University of Lagos NITDA Hub across academia, social impact initiatives, and non-governmental organizations.
You have been both a trainee and a trainer. Having experience on both sides of the classroom, what advice do you have for developing purpose-driven data skills?
When I train others, I emphasize the importance of understanding why they want to learn. Without a clear end goal and solid objectives, it is difficult to put forth the effort required to excel. Knowing who you want to be and why acquiring a specific skill is critical to achieving that goal is an important starting point. I also advise trainees to be consistent, as I have seen the benefits firsthand as a student at the University of Lagos. I realized early on that the best way to achieve any goal is to begin preparing as soon as possible. This entails putting in the necessary effort from the start, as it is the only way to ensure the best results and a thorough understanding.
In addition, trainees must also understand their learning styles and tailor their study methods accordingly. People process information differently; some may grasp concepts in class, while others may require additional review to fully comprehend. Understanding their learning style allows them to level the playing field and achieve the best results.
With all of this under their belts, I recommend that trainees choose courses or programs that teach the fundamentals of even the most complex data concepts, such as those offered by the NITDA Hub. This provides them with a very useful context that will help them when they apply their skills and as technology changes in the future. After that, I recommend that they practice a lot because proficiency comes from dealing with real-world problems.
“5 Minutes with” series
These articles share the stories of people around the world leveraging data and AI to drive impact.