NCP-AIO practice questions
NVIDIA · NCP-AIO · 200 questions
Original practice questions for the NVIDIA-Certified Professional: AI Operations (NCP-AIO) exam, covering installation and deployment (BCM, Base View, Mission Control toolkit, Kubernetes/Slurm/Run:ai install, DOCA on BlueField DPUs, NGC and cloud VMI containers, AI storage), administration (Slurm, Run:ai, Kubernetes, Fleet Command, MIG, AI data center architecture), workload management (inference and training on K8s/Run:ai/Slurm, team resource allocation, DCGM/NVSM/nvidia-smi, NGC runtime deploy), and troubleshooting and optimization (Docker, fabric manager for NVLink/NVSwitch, BCM, Magnum IO, storage performance) — framed in civic and municipal scenarios such as city GPU clusters, public-health inference, 311 models, and transit operations.
This course contains the use of artificial intelligence.
About the NCP-AIO exam
- Time allowed
- 2 hours
- Questions
- 30 multiple-choice plus 3 hands-on lab exercises
- Languages
- English
- Format
- Online, proctored remotely; MCQ section plus live virtual machine lab, each 50% of score
Exam details published by the vendor, checked 6 November 2026. Vendors change fees and formats without notice — confirm on the vendor's own page before you book.
Practice Quizzes
Test your knowledge with standard 20-question practice sets.
Quiz 1
Quiz 2
Quiz 3
Quiz 4
Quiz 5
Quiz 6
Quiz 7
Quiz 8
Quiz 9
Quiz 10
Browse by Domain
Study specific topics at your own pace.
Installation and Deployment · 62 questions
- A city IT ops lead needs one surface that shows which municipal GPU clusters host overnight public-health batch jobs versus daytime 311 inference, without SSHing into each head node. Which approach best fits that need?
- A transit authority wants one pane for rack health across bus-depot edge boxes and the downtown training cluster. Staff debate Mission Control versus only Base View on a single BCM head node. What should they choose?
- County emergency management asks whether Mission Control replaces BCM for adding a flood-model compute node to the civic cluster. What distinction should ops make?
- City GPU ops need real-time node health and utilization while 311 classification jobs spike each morning. Which BCM capability should they open first?
- Public-health inference nodes show amber in Base View while training nodes stay green during the same window. What is the most accurate ops interpretation to pursue first?
- A water-utility analytics rack drops offline overnight. Before rolling a truck to reseat cables, what should civic ops check first?
- Parks department computer-vision jobs finish early, yet Base View still shows high memory on two nodes. What should ops conclude about those charts?
- Housing authority fine-tuning jobs contend with 311 inference for the same GPUs. How should city ops enforce fair shares through BCM?
- Transit ETA model training must wait until after evening-rush inference. How should that blackout window be encoded?
- County assessor imagery training oversubscribes a node category and starves other municipal teams. What BCM action addresses the allocation problem?
- Emergency-management surge models need GPUs immediately during a flood event, while low-priority parks jobs still hold capacity. What should ops apply via the BCM-managed scheduler?
- After a vendor GPU firmware advisory, city BCM must prepare nodes before Monday’s 311 peak. Which BCM action set is appropriate?
- Two water-analytics nodes diverge from the golden software image before the next SCADA model run. What is the correct remediation focus?
- Library edge boxes need a security patch, but Friday catalog-vision inference must stay up. How should BCM patching be planned?
- Transit depot nodes report mixed CUDA driver levels after a partial update. What should BCM ops do to restore consistency?
- Only public-health ML engineers should submit PHI-touching jobs, while parks CV students must not. How should BCM access be designed?
- A contractor supporting the 311 model needs to watch Base View but must not sync software images. Which BCM permission approach is appropriate?
- A school-district intern accidentally received cluster-wide reboot capability on the education GPU lab. What should admins do?
- Housing authority data scientists share one project login for BCM. Security requires named users with stronger authentication. What change aligns with BCM user administration?
- City BCM must keep RDMA paths healthy for public-health multi-node training across nodes, DPUs, and switches. Where should those network settings be configured and monitored?
- Transit InfiniBand fabric shows flaps after a depot switch change. What should civic ops check first?
- BlueField DPUs on county edge nodes must be ready before DOCA services go live. What BCM-focused step should precede DOCA bring-up?
- 311 inference nodes and training nodes need different VLAN and QoS treatment on the city cluster. How should that be applied?
- A water utility adds a storage VLAN and must update node network settings without breaking GPUDirect-oriented AI data paths. What is the right BCM-oriented approach?
- A flood-forecast job fails with exit code 1 across three nodes. Where should civic ops begin root-cause analysis?
- After UPS maintenance, two permitting-model nodes appear dark. How should civic ops use BCM to diagnose the outage before deciding re-image versus power checks?
- Public-health job backlog grows even though GPUs look idle. What BCM diagnosis should civic ops pursue next?
- Transit ETA jobs fail only on category B nodes. How should civic ops use BCM to isolate the fault?
- The courts OCR queue stalls and BCM shows the head node’s disk is full for scheduler state. What should civic ops do?
- A city wants A100 inference nodes kept separate from H100 training nodes for 311 versus research workloads. What BCM capability should ops use?
- Parks computer-vision jobs and public-health training must not land on the same small-MIG pool. How should BCM enforce separation?
- A county adds L40S nodes for edge VLM pilots. What should ops do in BCM before admitting jobs to the new hardware?
- A school-district GPU lab needs some nodes reserved for district operations and others available as preemptible student capacity. Which BCM approach fits?
- The CIO asks for monthly GPU utilization figures for the smart-city AI budget. Where should civic ops generate usage, performance, and issue summaries?
- After a 311 outage, auditors ask for after-action records. How should civic ops preserve the story in BCM?
- A municipal AI platform team needs Kubernetes on NVIDIA GPU hosts. Which install path best matches NCP-AIO Domain 1 practice?
- After a BCM Kubernetes install, what should civic ops verify before exposing the first public-health inference Service?
- County edge depot sites must match downtown Kubernetes control-plane versions. What install approach should civic ops use?
- A school-district AI lab’s first pods will not schedule GPUs after a BCM Kubernetes install. What incomplete step should ops suspect?
- A city wants BlueField DPU-Arm services to accelerate east-west traffic for multi-node health models. What should ops deploy?
- Transit wants DPU telemetry in place before Run:ai edge workers join depot nodes. Which deployment step should come first?
- Water SCADA analytics hosts show BlueField DPUs up, but DOCA services are missing. What should civic ops correct?
- Emergency-management designs need DOCA on DPU-Arm for storage and network services without changing host GPU MIG layouts. What distinction should architects keep?
- A city platform wants Kubernetes AI workload governance for 311 versus research queues. Which Domain 1 install action fits?
- After Run:ai install, housing-authority projects cannot see GPUs. What should civic ops validate?
- A county is sequencing Run:ai onto a BCM-initialized Kubernetes GPU cluster. When should Run:ai be installed?
- A public-health training farm needs multi-node fine-tuning under BCM. Which scheduler install path matches Domain 1?
- Transit keeps overnight training on Slurm while Kubernetes serves daytime inference. What Domain 1 stance fits under BCM ops?
- After a school-district’s first BCM Slurm install, student jobs see CPUs only. What incomplete configuration should ops suspect?
- A new city AI rack’s first operational step is standing up cluster management. What should civic ops install and configure first?
- Civic NOC wants 311 inference nodes to keep running if one BCM head fails. What should ops configure?
- A county still runs a legacy Bright-style footprint and wants production models soon. What should civic ops finish first?
- An air-gapped city cluster needs a Triton image for 311 inference. How should civic ops deploy from NGC?
- Public-health GPU training needs a supported stack. Which container choice should civic ops make?
- A library CV pilot fails after pulling an NGC container. What mismatch should civic ops check first?
- County assessor imagery burst-trains in cloud while keeping parity with on-prem BCM nodes. What should ops deploy?
- On-prem GPUs are saturated during a wildfire surge. What should emergency management use for extra capacity?
- Transit ops must keep NVIDIA stack parity between depot nodes and cloud burst hosts. What should they prefer?
- City sizes storage for multi-node public-health training checkpoints. What must civic design emphasize beyond raw TB?
- 311 embeddings need fast reads while old transcripts can sit cold. How should civic ops tier storage?
- Water-utility training starves reading large sensor corpora over plain NFS. What storage path should civic design consider?
- Courts OCR farm needs durable datasets, fast scratch, and a model registry. What storage layout should civic ops follow?
Administration · 46 questions
- Library branch cameras run Metropolis-style apps at the edge. How should city ops administer them?
- Transit wants remote updates of depot occupancy models on edge boxes. What Fleet Command step comes first?
- A bad model push hits public-health clinic kiosks. What should civic ops use in Fleet Command?
- Parks CV gateways stop answering. What should civic ops check in Fleet Command before rolling trucks?
- County wants court lobby edges and library edges to limit shared blast radius. What should Fleet Command structure reflect?
- Emergency-management trailers need overnight model health checks without a tech on site. What should ops use?
- School-district AI lab edges must not share the same Fleet Command org as PHI clinic edges. Why?
- City is about to ship 311 lobby kiosks to a new site. What Fleet Command work should precede shipping?
- Public-health PHI training must not share Slurm QoS with open research. What should civic admins configure?
- After a transit node swap, Slurm schedules CPUs but no GPUs. What should civic admins correct?
- Housing fine-tune jobs need whole nodes without noisy neighbors. What Slurm administration action fits?
- County assessor array jobs flood the Slurm controller. What should civic admins do?
- Clerk OCR needs Slurm maintenance, but 311 Kubernetes inference must keep running on other nodes. What should admins do?
- School-district thesis jobs starve while weekend scrapers run. What Slurm admin change favors thesis work?
- Water-utility multi-node GPU jobs land on poorly connected hosts. What should Slurm admins configure?
- Emergency AI must jump ahead of batch library jobs during a surge. What Slurm administration fits?
- City wants Slurm jobs to mount civic datasets safely at start and clean up at end. What practice fits?
- Library consortium sites want shared weekend catalog-vision training. What Slurm arrangement should admins use?
- City plans dense health-model training with heavy all-reduce. What data center architecture choice fits?
- 311 inference wants many smaller nodes; overnight training wants dense NVLink domains. What should architecture reflect?
- Transit ETA training checkpoints crawl across the network. What architecture change helps?
- County power and cooling cannot add another full GPU rack. What architecture move should come first?
- Public-health PHI models must not share the same zone as open-data 311 models. What architecture fits?
- Emergency management wants on-prem primary AI plus cloud burst. What should architecture define?
- Library branches need light CV inference; heavy training stays downtown. What architecture split fits?
- Water SCADA analytics GPUs share a ToR with guest Wi-Fi. What architecture change is required?
- Courts OCR GPU row lost power in a prior UPS event. What architecture improvement fits?
- School-district teaching labs keep crashing production district models. What architecture control helps?
- City departments 311, public health, and transit share one Run:ai cluster. What admin setup fits?
- Housing researchers need Run:ai notebooks but leave GPUs idle for hours. What should admins configure?
- Transit inference cannot wait in a general training queue. What Run:ai administration fits?
- County wants training favored on weekends and inference on weekdays in Run:ai. What should admins tune?
- Public-health PHI projects are about to go live on Run:ai. What identity step should admins complete first?
- Parks CV needs a weekend festival quota bump without permanent entitlement creep. What should Run:ai admins do?
- School-district staff keep creating cluster-wide Run:ai roles. What should admins restrict?
- Emergency-management drills must preempt low-priority library jobs in Run:ai. What should admins configure?
- After maintenance, Run:ai still schedules to a cordoned civic GPU node. What should admins do?
- City platform admins run weekly Kubernetes GPU health checks. What should those checks include?
- City 311 inference runs as a Kubernetes Deployment. What availability control should admins apply before draining nodes?
- County Kubernetes hosts both public-health and open 311 namespaces. What should admins use to keep PHI traffic isolated?
- Transit ops, planning, and safety each run GPU jobs on one Kubernetes cluster. How should admins separate their capacity?
- Library GPU nodes need a controlled Kubernetes upgrade. What node-lifecycle practice should admins follow?
- City wants several 311 microservices to share A100 GPUs without starving each other. What should admins configure?
- Public-health overnight fine-tuning needs full GPUs. How should admins handle MIG on that BCM node category?
- A school-district GPU lab wants about twenty students to run homework models at once. What MIG approach fits?
- Transit finds mixed MIG and non-MIG GPU nodes confuse the scheduler. What should admins do?
Workload Management · 46 questions
- City needs a 311 text-classifier inference service on Kubernetes. How should the workload be deployed?
- A public-health symptom triage model will run on Kubernetes. How should its network exposure be deployed?
- Transit ETA inference must survive a single depot GPU node failure. What Kubernetes deployment pattern fits?
- Library branch kiosks need continuous catalog-vision inference. Which Kubernetes workload type should ops deploy?
- Water leak-detection inference is getting a new model. How should the Kubernetes rollout start?
- Courts OCR online inference must land on a specific GPU product class. What should the Kubernetes deploy request?
- Emergency chat-assist inference is quiet off-shift. How should Kubernetes deployment save civic GPU budget?
- Housing authority needs a policy-Q&A model served with predictable GPUs. How should inference be deployed?
- 311 Run:ai inference must stay responsive during Monday morning ticket storms. What should ops configure?
- Parks computer-vision inference should consume only parks capacity. How should the Run:ai workload be deployed?
- County assessor needs a new imagery model version in production. How should inference roll out?
- School-district demo day apps need temporary Run:ai inference access. What deployment control fits?
- Transit platform policy must stay centralized for ETA inference serving. What deployment approach should ops choose?
- Public-health needs a multi-node fine-tune on the civic Slurm cluster. How should the training workload be deployed?
- Transit overnight ETA training must process each weekly dataset chunk. Which Slurm pattern should ops deploy?
- Assessor imagery training must avoid MIG-sharing interference. How should the Slurm workload request capacity?
- Water-utility hyperparameter sweeps need many related Slurm training jobs on shared data. What packaging fits?
- Courts OCR model training crawls when the Slurm working directory sits on slow NFS. What placement fix fits?
- Emergency-management needs a surge retrain during a city drill. How should the Slurm training job be deployed?
- A library consortium must train a shared embedding model across two campus GPU sites. What Slurm approach fits?
- City data science needs distributed 311 model training on the municipal Run:ai platform. How should the workload be deployed?
- A public-health Run:ai training job requests more GPUs than the project quota allows. What should operators expect?
- Housing-authority fine-tunes must shrink when inference priority rises. What Run:ai training pattern fits?
- School-district teachers need to launch GPU training from notebooks without reinventing cluster settings. What fits?
- Transit ops compare Slurm and Run:ai for the same ETA training job on a Kubernetes-centric platform. Which choice fits?
- Parks needs a seasonal computer-vision retrain on a predictable cadence. How should Run:ai training be scheduled?
- City NOC suspects ECC errors on a 311 GPU node before escalating a ticket. What first-line check fits?
- Public-health training slowdowns have no clear application bug. What system tool triage fits?
- DGX-class civic systems should warn about fan and PSU problems before models fail loudly. What tool fits?
- Transit sees Run:ai jobs pending while teams claim the code is broken. What correlation helps capacity decisions?
- Library kiosk GPUs appear slow and nvidia-smi reports thermal throttle. What should ops conclude first?
- County ops want weekly GPU health trends in the municipal observability stack. What integration fits?
- Water-utility ops need GPU diagnostics on one suspect node without taking the whole partition down. What fits?
- School-lab GPUs disappear from nvidia-smi after a driver change. What should ops do before reimaging?
- After budget hearings, the city must split GPUs across 311, health, and transit on Run:ai. What allocation fits?
- Housing under-uses Slurm GPUs while parks over-uses. How should civic team allocation be adjusted?
- A student namespace on the school-district cluster risks claiming all GPUs. What Kubernetes control fits?
- An emergency drill needs temporary GPUs moved from libraries to emergency-management pools. What fits?
- Assessor and clerk teams keep over-promising GPUs. What cross-team allocation practice fits?
- 311 Run:ai burst must not starve health training for an entire quarter. What allocation balance fits?
- A mixed Slurm and Kubernetes civic site must allocate nodes to teams without double-booking. What practice fits?
- After a CVE patch, 311 serving must pick up an updated NGC TensorRT container at runtime. What fits?
- Public-health training must start from an NGC PyTorch container matched to the cluster CUDA level. What should ops pick?
- Transit wants GPU workload base images that stay supportable and reject unsigned sidecars. What practice fits?
- City must expand 311 capacity mid-quarter by adding a new GPU node to the existing BCM cluster. What fits?
- After hardware delivery, county must finish BCM provisioning so the node joins the right category and scheduler. What should ops do?
Troubleshooting and Optimization · 46 questions
- 311 inference pods CrashLoop after a toolkit misconfig and Docker cannot see GPUs. What should ops troubleshoot first?
- Public-health build cache fills disk and civic nodes run out of space. What Docker remediation fits?
- Transit containers cannot pull because air-gap registry mirrors on the Docker daemon are wrong. What should ops fix?
- Library CV container runs on CPU only because GPU request flags were omitted in the unit file. What fix fits?
- Courts OCR workers die with permission errors on NVIDIA device nodes. What Docker troubleshooting step fits?
- Water-utility hosts run an old Docker engine that rejects the nvidia runtime required by the new stack. What should ops recognize?
- School-lab containers inherit the wrong default runtime after a partial toolkit install. What should ops repair?
- Emergency container logs show OOM kills after Docker memory limits were copied from CPU-only services. What should ops do?
- City nodes have duplicate Docker and containerd paths and Kubernetes calls the wrong runtime. What should ops untangle?
- A dense civic training node shows GPUs isolated after maintenance because fabric manager is stopped. What should ops do?
- Public-health all-reduce hangs and topology tools show a broken NVSwitch path while fabric manager logs confirm issues. What triage fits?
- Transit dense GPU node fails fabric manager on boot after a BIOS change. What should ops do?
- County training slowdown correlates with fabric manager in a degraded state, not application code. What should ops conclude?
- Library mistakenly installs fabric manager on non-NVSwitch nodes and creates service noise. What should ops know?
- Water multi-GPU job sees peer-to-peer disabled; fabric manager health is the fix path. What should ops do?
- Emergency retrain on an NVSwitch rack fails until fabric manager matches the driver branch. What alignment fits?
- School-district dense box shows intermittent NCCL errors while fabric manager crash-loops overnight. What should ops stabilize?
- City monitoring alerts that fabric manager is inactive after image sync omitted enabling the unit. What should ops ensure?
- Assessor multi-GPU throughput halves while fabric manager reports partial NVSwitch links down. What should ops interpret?
- BCM head node disk fills with an old monitoring DB and municipal Base View freezes. What troubleshooting fits?
- A new water node provisions but never phones home because BCM network boot and DHCP settings are wrong. What should ops fix?
- City image sync reports success, but GPU nodes diverge after reboot. BCM software images do not match node categories. What should ops fix?
- After BCM head-node failover, Slurm job submits through BCM fail for the transit authority. What should ops repair?
- Public-health BCM alerts flood the NOC with non-actionable noise. What should ops tune?
- Transit operators cannot open Base View after the BCM GUI certificate expired. What fixes access?
- County BCM user authentication fails after directory sync; operators lose expected roles. What should ops repair?
- School-lab GPU nodes show as unknown in BCM after motherboard swaps. What reconciles identity?
- Emergency-management BCM admin commands hang cluster-wide. A stuck report job holds a database lock. What clears the freeze?
- City multi-GPU training stalls on IO. Magnum IO GPUDirect Storage to the parallel filesystem is misconfigured. What should ops fix?
- Public-health multi-node NCCL over InfiniBand is unexpectedly slow. What should ops verify in the Magnum IO stack?
- Transit checkpoint writes bypass GPUDirect and overload CPUs. What should ops check?
- Library CV containers show broken GPUDirect Storage after an image update. Magnum IO libraries no longer match. What restores GDS?
- Water-utility UCX and NCCL environment variables conflict with Magnum IO guidance after a toolkit bump. What should ops do?
- County storage disables a mount option required for GPUDirect Storage; Magnum IO falls back silently and training slows. What should ops recognize?
- Emergency multi-node jobs regress after an OFED update. What should ops consult before changing more packages?
- School-district single-node jobs are fine, but multi-node IO is broken. What Magnum IO area should ops troubleshoot?
- 311 feature-store ingest onto GPU nodes should use Magnum IO. Ops must prove the data path is healthy end to end. What fits?
- Assessor training GPUs sit idle while jobs wait on NFS metadata. What is the primary bottleneck to diagnose?
- Public-health checkpoints on spinning disks stall training for hours. What storage remediation fits?
- Transit parallel filesystem shows imbalance across OSTs and ETA training throughput drops. What restores balance?
- Library computer-vision datasets on object storage thrash with tiny reads. What improves the access pattern?
- Courts OCR batches compete with 311 embedding rebuilds on the same LUN and both slow down. What should ops apply?
- Water SCADA analytics nodes report GPUDirect Storage errors under load. What should ops validate?
- City NGC pulls fail authentication against the municipal mirror, blocking a Triton image deploy. What should ops fix?
- A public-health NGC container starts then exits because CUDA user-mode libraries are missing on the node. What closes the toolkit gap?
- Transit copied an NGC edge ARM tag onto x86 server nodes and the container will not run. What mismatch should ops catch?
These questions are original practice material and are NOT actual exam questions or brain-dump content. All vendor marks are trademarks of their respective owners. This site is not affiliated with, endorsed by, or sponsored by the exam vendor.