Saurabh Gupta, CEO of The Modern Data Company, made a point in a recent interview that landed with us: the biggest risk in enterprise AI may be building yourself into a corner you cannot exit. His argument is that organisations are adopting AI platforms so tightly coupled to specific vendors that switching later becomes prohibitively expensive. The technical term is lock-in. The practical term is regret.
A Munich-based automotive supplier asked us to find a head of data architecture who could evaluate multi-cloud strategies. The requirement was explicit: any candidate who defaulted to a single hyperscaler’s tooling would be screened out. The company had spent two years unwinding a previous platform decision and was determined to avoid repeating it. That search took longer than expected because most candidates had deep expertise in one ecosystem and limited experience designing for portability.
The skill set companies now want sits at the intersection of technical fluency and strategic foresight. Data engineers who understand Kubernetes, open-source orchestration layers, and abstraction patterns are in demand precisely because they can build systems that survive a vendor renegotiation or a strategic pivot. The premium goes to people who can answer the question: what happens when we want to leave?
We have started asking candidates in final-round prep sessions how they would design infrastructure for eventual migration. The ones who can walk through container orchestration, API abstraction, and data-layer independence with specific examples are advancing faster. One senior architect we placed in Zurich last quarter won the role partly because she had led a migration away from a proprietary analytics platform and could describe exactly what made it painful and what she would do differently.
Prompted by reporting from BigDATAwire.