Interview prep for AI roles
Pick a company and practice the rounds its candidates report, with a worked answer for every case.
Every case
1.2Mclaims a year
Claims intake for an insurer
30 daysto pretrain 34B
A training cluster for a pharma research customer
800 msfirst token, p95
Serve a 70B coding assistant at 400 requests a second
140action codes, live
Live call assist on four GPUs
6,000agents on calls
Live suggestions for a telecom contact center
24,000authorizations a second
Score card payments inside a 40 ms budget
2 weeksbefore launch
Review a design doc: a maintenance agent for 38,000 apartments
16nodes, as fast as 12
Four new nodes and no speed-up
168files of 5 GB
Ship an 840 GB checkpoint to 1,536 hosts
Role guides
What the job posts ask for, the skills behind each case, and the labs that train them.
AI Solutions Architect
Applied AI Architect, Applied AI Engineer and Forward Deployed Engineer interviews.
Open the guideNVIDIA Solutions Architect
Solutions Architect interviews for AI factories, GPU clusters and networking.
Open the guideML and inference engineer
Inference, ML infrastructure and machine learning engineer interviews.
Open the guideJust want to practice ML?
Write the code behind machine learning, from your first NumPy array to the attention inside a GPT, in your browser.
Every case is our own, modeled on the formats candidates describe publicly. Preporato is not affiliated with Anthropic, OpenAI, NVIDIA, AWS, Palantir, CoreWeave or Microsoft. These cases help you prepare and don't promise an interview or an offer.