AI Insights

The Most Important Data Science Soft Skills (By a 20+ Years Experienced GIS & Remote Sensing Veteran)

January 30, 2022

article featured image

What are the most essential soft skills to work better in data science teams, solve problems more effectively, and excel in your career? Gijs van den Dool, an international consultant and researcher in remote sensing with 20+ years of experience shares his learnings. 

Can you describe your journey into AI/ data science?

Gijs career in short:

  • Many years ago started with GIS
  • Has a consultancy firm for environmental studies
  • 10 years in academia 
  • Worked in Insurance, built models for damage/non-damage (floods, etc.) 
  • Now independent researcher, applying AI

I have been interested in AI and Data Science for a long time, and I think this started during my BSc in International Land and Water Management (in Velp, the Netherlands), and has followed me throughout my career as a GeoSpatial Specialist. After graduation in 1998, I started as a junior GIS specialist in a large Dutch environmental consultancy firm in their surface and groundwater group, creating visualizations for groundwater quality studies and environmental assessment plans. The problem with (large) consultancy firms is that they have tight budgets, and there is not much time for Research and Development, when not related to a project or internally budgeted. 

I left the Netherland after a few years of making maps and not much progress, and settled, after a road trip, in one of the academic centres in Finland (Kuopio), where I worked at the University of Kuopio, the Geological Survey of Finland and the Finnish National Health institute. I mainly studied spatial problems, like routing intelligent household waste collection, designing smart sensors for collection bins, and creating a monitoring system for 3D reactive transport in contaminated aquifers. My last assignment was a predictive module for contamination detection in drinking water distribution systems. This was in the early 2000s, and IoT, Big Data, AI, and ML, were only just discussed, often this period is referred to as the revival of AI because of increased computing power, data storage efficiency, and refinements of machine learning / neural network algorithms.

After my research contract ended, and the research units started to work on topics out of my field of expertise, I moved to Paris (France) on an invitation to look into flooding occurrences and probability to assist the (re)insurance industry in their pricing policies and risk assessments. During this period, I obtained my MSc in GIScience and studied the possibility to assess/predict wild/forest fire occurrence using a GIS approach instead of pure statistical (mainly probability) methods. The use of AI/DL techniques is only recently, with the increased availability of satellite data, especially optical and SAR data, more integrated into the workflows to create risk models. Traditionally risk is expressed as a probability (Occurrence of Exceedance), but now also Parametric Solutions are accepted as a pay-out method, and it is in these solutions that AI plays a significant role.   

Currently, I am operating as an international consultant and researcher in remote sensing (earth observations), climate change, GeoSpatial Solutions, and Artificial Intelligence, and borrowing algorithms from both the ML and DL domains to solve (geo)spatial problems. 

What did you learn throughout your career, which can be beneficial to any data scientist and engineer?

Understand the problem

  • Take a step back
  • Ask “What would you like the data to tell you” 
  • Understand and value that team members have different perspectives

Before you should try to solve the question, you first need to understand the problem, what is it that is asked of me, because we are all having our unique points of view, experiences, and backgrounds so misunderstandings can happen very quickly. The best thing to do at the beginning of a project is to take a step back and try to see the problem from the person’s perspective who is asking the question. For example, in one of my assignments, we were asked to look into a method to assess flooding damage. We designed a beautiful, elegant, and sophisticated approach to predict the amount of damage, only to discover during one of the last project meetings that what they wanted to know was: do we get wet feet, and how often do we get wet feet.

We could have avoided many late meetings and many lost hours, trying to make the data fit our problem, while what we should have done is asked the right question at the right time and try to understand the situation better. 

Another thing I learned (the hard way) is that sometimes it is just not possible to make the data tell you the story you would like it to tell you. There are many reasons this can happen; perhaps the data is not good enough, or it is good enough but not in the precision you would like it to be, or you have excellent data, but there is not enough data. 

I see this often happening, mostly when we work on AI/ML problems, where we look at the standard error, the accuracy, and F1 scores, and judge based on these statistics that the data is useful. In the presentations, we forget to take that step back and look critical to the visual, checking how good the exhibit is, and asking: will my audience not understand what I show them on this slide? I think we should spend more time, when we are presenting our data, on the story, the narrative, behind the data, and not only the graph/map.

Again by listening, and understanding, there is (especially in GIS) more than one solution to a problem, and I like to quote George Box’s famous words: “all models are wrong, some are (more) useful (than others).”, and problem-solving is like building a model. So, most of the time, we are wrong and are not solving the problem, but we are getting close to an acceptable solution.

Improve your ability to listen 

  • Listen to the partner and collaborators 
  • Filter out and note the most important information in a discussion
  • Give people recognition by listening more and talking less 

The other thing I learned early is that it is essential to listen, and as a GEOScientist, this comes almost automatically because Geographical Information doesn’t exist without data, and before we (as GIScientist) can put the GI in a System, we should understand the data; where it came from, who created the data, why was this data created, is it static or dynamic, and how will this data be used, and if (re)used in what kind of type, format, or structure would be the most efficient to handle the requests. 

So, by active listening, this kind of information is filtered out of the conversation without asking too many technical questions, which means that the meetings are feeling natural and are much more beneficial to the project because the discussion is (again) on the problem you want to solve and not on the technical details. 

Also, as a GIScientist, you are often not the domain expert, but creating the relationships between the different domains, e.g. talking to both the remote sensing expert and the farmer while discussing a platform to forecast yield failure. I like to moderate these kinds of discussions, and often they lead to new insights when you allow all participants to tell their side of the story and recognize their expertise in their fields. The discussions I enjoy most are discussions where the conversation flows, and the group is working together towards a mutual understanding.

Can you share a point in your career where things got a bit difficult? And how did you overcome roadblocks?

I can try to give some mindset-related tips, and the first one should be to: 

Stay Positive, even when things are not looking so good, because the moment you give up, things start to get complicated and harder to overcome. 

I am starting late, as a consultant, and this is, you could say, my fourth career, or career change, and the least secure, but I am optimistic that I can succeed because if I don’t believe in this adventure, it will fail before it even started. Determination is another mindset, because a positive outlook alone will not get you where you think you should be, or want to be. And lastly, compassion, you have to feel it is worth doing because when things are difficult without these three “mindsets”, it will be tough to make the changes you think you should make. 

How did the Omdena experience help you grow your skills?

Build out my network 

I think the projects Omdena is running are unique, in the sense that, because they are not on the critical path for the organizations asking for help, the projects bring together a unique combination of skills, backgrounds, and experience, and it is in under these circumstances that most creativity will surface, and it is possible to test different approaches to solve a problem. So, while you are working in this environment, you get new ideas from the people around you, and you know who to ask when you run into a similar problem later, in the same project or the next.

I am not a programmer, by trade, so knowing who to call when I can’t solve a coding problem myself is one of the benefits of participating in an Omdena project. An Omdena project definitely brings you in contact with people you otherwise (most likely) wouldn’t have met, and worked within a project.

Make the best out of my time

One of the most important aspects of participating in an Omdema project, for me, is spending time with the teams and working on the problems. 

The solutions often require new techniques or new approaches, some I have not heard before, which is very stimulating. It keeps my knowledge current because, to deliver something useful at the end of the project, it is essential to understand (at least the basics) what teams are proposing and bring value to the discussions. Some of the knowledge, I can then implement in the consultancies that I am running. 

Teach and share 

The open discussions, in the meetings, are crucial to the success of an (Omdena) project, and I find it very motivating to be part of these meetings, not only to share the information and knowledge I have, but to hear how others have solved their problem, what their pain points were, and work collectively towards a solution. 

Sometimes that could mean that you have to step outside the project, run a one-off workshop, or do a one-on-one session; that is great and very helpful.  When you have to explain something, the thing you describe gets much clearer, not only for the audience but also for yourself, because teaching is learning.  I find it very valuable to have an environment like Omdena, where you have the freedom to do this of things, run experiments, and discover new ways of problem-solving.

If you would start all over again in your career, what would you do differently?

That is a difficult question, mainly because of the choices I made, I am where I am at the moment; an independent researcher and consultant in GeoSpatial problems using AI, EO, and working in a stimulating environment. There are a few regrets, some steps I could have skipped (in hindsight), and I could have arrived earlier at this point, but then I would have missed out on some other learning moments. I think that the thing I regret most is that I didn’t get my MSc earlier, but then circumstances didn’t allow me to get an MSc earlier, so if I have to give some advice: never stop dreaming, learning, and growing, and don’t start feeling too comfortable, because at least for me that are the moments I didn’t make progress I wanted to make – because it was too comfortable where I was at those times. 

Any closing words?

That you are never too old to learn, and that you should not stop developing when you feel comfortable (deep in the rabbits’ fur; Jostein Gaarder, Sophie’s World), there are always new things to discover, experience, and master.

Ready to test your skills?

If you’re interested in collaborating, apply to join an Omdena project at:

media card
Filling a Gap in the Iraq AI Sector and Launching my Own Startup – by Mohammed Zuhair, Ph.D
media card
Why Collaborative AI Projects Beat Competitions and How it Helped me to Get a Job as a Data Analytics Consultant
media card
From a Junior Machine Learning Engineer to an Associate Data Engineer in Only 12 Months
media card
Falling in Love with Data Science and How I Got a Job Offer as Data Scientist at Insight