Humans are notoriously poor at predicting the future (especially in the longer term). However, it still can be a useful exercise. Let’s look at what I think 2021 will bring to data science, engineering, and strategy. This list is in no particular order and is mostly focused on my observations in Germany:
- Roles stabilizing. In the last years, new roles in data have popped up, such as Machine Learning Engineer, Data Product Owner, and others. Those haven’t entirely caught on yet, with larger companies still resorting to the traditional role of Data Scientist. I would argue it is a good idea to have more positions available, with their associated skillsets - both for employees and employers.
- Consolidation of MLOps tools. Last year was the year where MLOPs tools and startups exploded in variety. Still, many such companies offer just a variation of the same product. 2021 will see many weeded out, while the frontrunners capture most of the market.
- DataOps growing in hype. The application of agile and lean methodologies in data science and engineering will become more widely discussed.
- Data Strategy goes mainstream. I have been covering this topic for the last two years, yet this job description still remains rare, especially in Europe. This does not, however, mean that the role is not filled by existing people. They might just work under a different tile. Still, this field is essential for successfully delivering data projects, and I’m optimistic for further growth and acceptance.
- xAI in production. xAI has been climbing the Gartner Hype Curve for a long time and now reaches the point where it starts to deliver results. There are still roadblocks to this subfield’s success (such as unstable open-source and arcane skillsets required). Still, new tools are emerging to bring it forward in production.
- Further data engineering explosion. No data science project is successful without its foundation - data engineering. Companies have been late in recognizing this, and there will still be catching up in 2021.
- Smart data cleaning and ETL tools. Everybody is aware of the time spent cleaning data. This has been a tricky problem to solve so far with tools, but new developments such as Cloud Data Prep will spawn competitors.
With this list in mind, I want to make two wishes for 2021 in data. First, I hope the whole field, but especially the ML part of it, becomes more “boring”, but useful. Second, we start to use this fantastic technology to solve the pressing issues we face and move to a more optimistic and ambitious future.