When Senior Machine Learning Engineer Andrea Scholtz first developed an interest in all things AI related, he didn’t anticipate the incredibly rewarding journey ahead. Driven by curiosity and a passion for tackling real-world challenges, his journey led him from applied mathematician to working at a machine learning start-up and then finding his place at JUMO. He shares his passion for harnessing technology to serve the financial sector better.
‘I had the remarkable opportunity to work on migrating an on-premise system for training a large recurrent neural network to a cloud-based framework on Microsoft Azure, early in my career. This was revolutionary for its time, and for me, as most South African financial institutions hadn’t yet embraced cloud-based computing or machine learning models. Seeing the potential, I transitioned into the fintech industry and joined the Prediction Services team at JUMO. My role centered on implementing automated model deployment pipelines, developing scalable training services, and establishing a model monitoring system. By automating these processes, we reduced deployment times from days to hours and significantly increased model production.
Later, I moved to the Data Science and Analytics team, where I currently work on building tools to enhance model training and improve predictions on customer defaults. The role of AI in our work cannot be overstated. Our models determine a customer’s propensity to repay a loan, facilitating responsible digital lending.
The complexity of the systems integration is immense, but the reward – unlocking credit for those who need it – is well worth the challenge.
In underserved African communities, data collection presents one of the biggest obstacles. We need to creatively utilise the data at our disposal to accurately understand a customer’s financial situation. It’s here that our partnership with mobile network operators enables the data insights we develop to support our mission to serve financially excluded customer segments.
AI’s remarkable capabilities have allowed us to expand our operations and extend our reach to communities in need of finance, while fostering economic growth in underdeveloped areas. Our ultimate goal is to provide tailor-made solutions that sustainably benefit people and business.
In an industry like ours, being forward-thinking is imperative. We constantly look towards emerging trends in AI that can further our mission. For me, one area of interest is AI ethics and regulation. The potential for AI to revolutionise our lives, granting us more time for creative pursuits, is undeniable. Nevertheless, we need to remain acutely aware of the potential for misuse. The advent of AI ethics poses a crucial challenge, urging us to scrutinise the applications of AI across various sectors, including the financial industry. Ensuring the explainability of our models and how they serve all individuals equitably is paramount. This approach aligns perfectly with our objective of improving lives through access to financial services.
Looking ahead, I’m anticipating a continuous evolution of the use of AI at JUMO and in the fintech industry as a whole, as we develop multifaceted systems that balance automation and personalisation. Automation allows us to scale our operations effectively, while personalisation ensures customers receive the tailored services they deserve.
My own AI and fintech journey has certainly been one of continuous learning, growth, and profound impact. The challenges we face – data collection, model complexity, and ethical dilemmas – only serve to fuel our innovation and commitment to our mission. We want to shape a more inclusive financial future for Africa.
Connect with Andrea on Linkedin
Disclaimer: Some of this blog was predictably created using generative AI tool ChatGPT. While it crafted the copy mostly sensibly using Andrea’s answers to interview-style questions, some statements came out vastly overinflated and still required human intervention by way of editing. It is, however, remarkable how seamlessly their voices intertwined, blurring the lines between human and machine for a more efficient and innovative way of doing things.