Interview with Machine Learning Specialist Marthinus Bosman
Machine Learning Specialist and Speaker at Swiss Testing Day 2020, Marthinus Bosman, shares details about his background and interests. Find out more in the interview.
Marthinus, can you pinpoint one moment or person that was instrumental in your decision to pick the career path as a Machine Learning Specialist?
Whilst studying I attended the deep learning indaba at Stellenbosch University, it was at this event where I was able to attend a talk from Jeff Dean, Senior Fellow at Google AI, Google Brain lead and co-founder and co-designer of TensorFlow. He gave a talk on Real Life Machine Learning and to me it was the first time I was able to see truly practical and innovative applications for Machine Learning beyond the normal Speech and Vision use cases. He also gave insight how this field is especially growing in African countries. I think that talk was what convinced me that there is still a large gap to fill for Machine Learning applications and is what made me decide to try and find those gaps.
What is the most interesting aspect of your job?
I think Machine Learning as a whole is becoming far more accessible to use by anyone without any specific technical knowledge, but the part that’s always the most interesting, and often challenging, is finding ways to represent problems as Machine Learning problems, which usually entails manipulating data in a creative way, or finding ways to generate useful data. I still can’t find a formula for this, and it’s always fascinating working in a team, where a team member is able to recognize a solution in a way I completely missed.
And the most challenging aspect?
Getting access to data or systems, and then making sense of where it fits in. Every time, that is the part that takes the longest and causes the most frustration. As a company I think solving that struggle is what we’ve spent the most time on and where we’ve grown the most. Luckily, I can see that other companies have started getting a better understanding of how we go about understanding systems through data, and we’ve gotten a lot better at communicating our needs.
What AI/ML trends are you most interested in?
I’m always more interested in niche ML uses. The larger trend these days are building more and more massive systems to get the last bit of accuracy on something like Machine Vision or Speech. Personally, I’m not that interested in Machine Learning becoming better in things we’re already great at. Machine Learning, to me, is an efficient way to find patterns or predict future behavior of systems that are too complex for humans to do. I’m interested in the new types of problems that Machine Learning are able to solve that humans can’t, and beyond that the better understanding we’re able to derive on those problems based on the Machine Learning output.
You’ll be speaking at Swiss Testing Day about machine learning supported advanced analytics for a shift-right testing environment. Why did you choose this topic?
Like I said for the previous question, I’m always interested in problems that Machine Learning can help solve that are complex for people. I think today, companies spend a massive amount of money and thousands of man-hours on trying to ensure that the software they publish is performing as expected, and it’s a problem far from being solved. As a problem, it’s also not necessarily clear how Machine Learning can be used to help. That’s why it’s an interesting topic for me to get in to, as I think there’s great potential for Machine Learning to help improve as well as give better insight into the problem.
What do you enjoy doing outside of the office?
Growing up a block away from the ocean I spent a lot of time on it. Whether it’s to go catch crayfish, take a trip around Table Mountain or to go waterskiing, I’m either on or behind a boat. When it gets colder, which in South Africa is closer to 10°C as opposed to what the Swiss would consider cold, I try and go camping as often as I can.