Ian Hansel is the Director of the AI company Verge Labs, and is an experience Data Scientist and Data Science Instructor.

In this interview, he shares some fantastic insights and advice on getting started in Data Science and building a successful career.

As a Director of Verge Labs, can you please tell us a little about your role, what a typical day may look like for you, and what you enjoy most about it?

Verge Labs is a company I established with my co-founders Anthony Tockar and Tim Garnsey to bridge the gap between cutting-edge academic research and outcome-centric business innovation. We offer a subscription service to provide practical research and software tools, as well as time to help hire and build teams, and implement these tools on specific business problems. My role is to assist companies getting started with Machine Learning projects that adds value to their business and develop the utilities in the subscription service.

In terms of a typical day it will be some mixture of the following:

• Advising companies on how to implement Machine Learning that adds real value to their business;
• Developing applications that we have identified businesses require from our consulting;
• Talking about Machine Learning and evangelising the field in general at events like meetups or conferences or through our blog; and
• Drinking too much coffee.

During your time as a Data Scientist, you would have worked on some interesting projects. What are some of the successes that you are most proud of you can talk about at a high level?

Some recent projects have been working in the space of geo-spatial analysis. I’ve found that to be very interesting as there’s some very rich information that can be put to good use with some powerful geo-spatial data analysis and visualisation tools.

I’ve been able to combine some very interesting insights from location data such as traffic counts, ABS census data, points of interest from Google Maps API and weather data to describe why some brick and mortar retail stores perform well (or poorly). Then extrapolating that to find locations that might be suitable for opening up a new store. The client previously had a very manual process for evaluating locations, I was able to build a system that assessed many locations all across Australia and explain why they might be suitable targets for a new location.

What are some of the biggest challenges you’ve faced throughout your career, and how have you overcome them?

Communication is a big one.

Finding the right level to communicate ideas is crucial. There will be people that want to go deep into the details but sometimes people will want to understand only the outcomes. Other times they might want to have some intuition about how a model works and what it’s doing, but not dive into any of the mathematics underneath.

Finding the right way to get the message across for a customer can be challenging but in many ways just as important (if not more) as the underlying analytics work. One way I’ve been able to develop my communication skills is through teaching Data Science courses.

Keeping current is a another challenge but also one of the more enjoyable aspects of the job.

It’s not about jumping on the newest and shiniest trend, but identifying what’s going to add value for a client and make you a better data scientist for having that skill in your toolkit.

You were also a Data Science Instructor at General Assembly. Can you tell us a bit about some of the courses you taught, the main challenges faced by students, and what some of the most popular subjects were?

Yes, I really enjoy teaching. One of main pillars of Data Science is Pedagogy and I’ve found it helps a lot with my own communication skills. I’ve taken great pride in working with students after the course is over and am very happy to see students going on the land some great roles in analytics within some amazing organisations, even forming their own companies in some cases.

The course I taught was the part time Data Science class. It covered a range of topics that are a great jumping on point, with a focus on being very applied. Going over everything in Introduction to Statistical Learning (James, Witten, Hastie and Tibshirani), some Python programming and basics of Cloud Computing. We also covered some other skills like how to structure a data science project and communicate it to stakeholders.

I like to bring in people from industry so the students can hear from different people, I find there are lots of different backgrounds in the field and it’s great to hear from a diverse group of people about their journey. In terms of most popular subjects, everyone has a different favourite which is great because everyone gets something different from the course.

The main challenge is understanding some of the more complex material. I see big part of the job is to encourage students to keep going and keep working on things until you find a mental model of a concept that works for you. My challenge is to break down concepts into something that is relatable to a student. (A skill that’s very useful in business as well).

What did you enjoy most about consulting, and what are your thoughts on aspiring and junior Data Scientists working as consultants when trying to grow their careers?

I enjoy facing new problems on a regular basis. The great statistician, John Tukey said, ”The best thing about being a statistician is that you get to play in everyone’s backyard.’ I find that to be very true, and the great thing about it is once you get exposure to many problems you get better at breaking them down into something that’s solvable with Machine Learning.

I think working as a consultant when you are a junior Data Scientist is a good way to learn things like; time management, displaying value, working on the right problem and problem decomposition. One thing to be aware of is you need to find the right group to consult with, as a junior that can be difficult to identify. To help you find the right group to work with I’d recommend finding a mentor who’s been there and done that.

What areas of research are of particular interest to you?

The most interesting (and broad) areas of research to me is anything with human-in-the-loop. I’m interested in ways people are considering how to leverage people and algorithms together, rather than seeing it as mass-automation and a zero-sum game. These things include; interfaces to data and models (with particular consideration to UI/UX), understanding how Machine Learning models are working, and incorporating feedback into a Machine Learning system. For a really good overview I’d recommend looking at what comes out of the Human-in-the-loop Data Analysis (HILDA) conference and also having a look at the University of Washington Interactive Data Lab.

How have you seen the field of Data Science change throughout your career, and what advances, such as in the areas of AI and Deep Learning, excite you the most?

I think the main thing that’s changed is accessibility. I am blown away by how many amazing free resources there are for Data Science right now. Everything from learning materials with MOOCs to being able to run and try out code and build machine learning models on virtual machines for free. The level of abstraction around Machine Learning API’s is making it very easy to get started which is a double-edged sword, but another conversation. And in the future things like AutoML will make it even easier.

In terms of technologies like Deep Learning they are certainly impressive and I believe there are ideas that are being developed that will enable some very interesting types of analysis. Again, one of the major things that’s happened in the field of Deep Learning over the last few years is accessibility. Looking at previous implementations on something like Theano or Cuda and comparing it now to Keras or Tensorflow you can see how easy it’s become to build these types of models (which can solve some very interesting problems). Not only that but deploying these models into many different environments like mobile phones or self-driving cars is becoming easier which is increasing the actual usage of Machine Learning in the community.

What are some tips you’d like to share with aspiring and junior Data Scientists looking to enter the field and develop their careers?

I don’t really have any specific technical advice, like go learn about Recurrent Neural Networks – even though that stuff is important. The reason for that is I believe a Data Scientist should always be learning and at the same time it’s impossible to know everything, there are some things you’ll need to learn as it becomes important to a project. You have to know the basics and have a strong foundation, then be comfortable building upon that knowledge.

That said my three tips for those getting started would be:

1. Find a mentor: I’ve been fortunate enough to have many people help me with my career, in particular Eugene Dubossarsky. To find a good mentor you need to show that you’re passionate about the field and are willing to take on advice. I’ve found the data science community to be very welcoming to newcomers I’d and encourage anyone to go along to a few meetups and meet people in the field;
2. Find a clique: Similarly to having a mentor, it’s good to have a group of people at a similar level to yourself to bounce ideas off and also keep up to date with what’s happening in the field. It will also help out later as you develop in the field to know other people in different companies. Knowing the person you end up working with on a project helps cut through the minutia; and
3. Keep at it: When you’re getting started it can be a lot to take in. But it’s crucial that you understand the basic concepts inside-out. It can be discouraging when things like models or projects don’t go the way you’d like but as long as you are learning from them and not making the same mistake twice you’re heading in the right direction.

I just wanted to add that if you’re in a company that’s looking to implement machine learning and artificial intelligence, feel free to get in touch and we’d be happy to talk to you. Also, if you’re an individual looking to get into the field please feel free to reach out to me directly. I’ve been fortunate enough to have people be gracious with their time early in my career and always appreciate a chance to ‘pay it forward’ 🙂

Disclaimer: The opinions expressed by Ian represent his own personal views and not those of his employer.