NVIDIA interview prep
Solutions Architect interviews for AI infrastructure, and ML engineering.
NVIDIA publishes no interview process that we could find. Candidates report a recruiter or hiring manager call, technical calls with Solutions Architects, then a loop of several interviews.
The rounds
As candidates report them. Rounds and their order vary by team, so check with your recruiter.
Practice caseNo case yet
Screens
Final rounds
Screens
- Coding roundsEasy to mediumData structures and clean codeNo case yet
- Serving screenOn your own workDepth on the CUDA and serving you didNo case yet
Final rounds
Cases
Same role, other companies
These rounds come up at other companies hiring for the same work.
800 msfirst token, p95
Serve a 70B coding assistant at 400 requests a second
24,000authorizations a second
Score card payments inside a 40 ms budget
168files of 5 GB
Ship an 840 GB checkpoint to 1,536 hosts
3 weeksto launch
Review a teammate's design doc for a streaming API
GQAwith RoPE, in NumPy
Fix the attention layer, then add a KV cache
Same seedsame text
A reproducible sampler with top-k, top-p and speculative acceptance
3 partsearlier tests keep passing
Build a continuous-batching scheduler with paged KV memory
5 daysto send it
Take-home: make a slow inference server fast
Each case's source tag says which company and role reported its format, how many independent sources describe it, and when. It doesn't mean a company uses our question. Every scenario and number is ours. Preporato is not affiliated with NVIDIA. These cases help you prepare and don't promise an interview or an offer.