What to expect
Coding rounds
Live coding in Python or Go with tests and edge cases, sometimes implementing ML routines such as token decoding from scratch.
ML and system design
A deep dive on eval suites, fine-tuning, retrieval-augmented generation, or multi-tenant inference with real tradeoffs and cost thinking.
Behavioral
Discussion of reliable infrastructure, async collaboration, and why Cohere and enterprise AI specifically, often with design-doc examples.