There are many interesting bits in the recent survey from Anaconda (“The State of Data Science 2020”) but one thing stood out for me: the difference between what Universities teach in their data science programs and what companies need. As I and many others have noted, while the media writes articles about data scientists, data engineering skills are actually more critical to most companies. So even in a survey aimed at teasing out data science needs, companies cite “big data management” as their most pressing need, and another quarter of respondents cited “Engineering Skills” as another pertinent skill. On the modeling side, Universities need to start incorporating practical deep learning courses into their curricula:
For many machine learning projects, the goal is to deploy models to production. Once again the survey reminds us that different personas have different blockers. Specific roadblocks highlight areas where a specific role or persona may not have adequate skills or experience.
- Data scientists list “managing dependencies” and “skills gap” – both pertaining to MLOps – as their main blockers.
- In comparison, developers cite “security” and the “need to re-code a model” as their main roadblocks
As companies deploy more models, managing and taming risks associated with ML will require machine learning platforms that can accommodate people from different backgrounds and disciplines. Legal, security, compliance teams will need to work side-by-side with your technical and product teams. ML platforms will increasingly need to enable collaboration between interdisciplinary and cross-functional teams.
Subscribe to our Newsletter:
We also publish a popular newsletter where we share highlights from recent episodes, trends in AI / machine learning / data, and a collection of recommendations.