GPU Utilization
87%
Active Workloads
24 / 31
Monthly Savings
$41.2K
Compliance Posture
100%
Cluster Utilization (24h)
CPU, GPU, and memory across all nodes
GPU
CPU
Priority Queue
Clinical > Research > Batch
CRITICAL
Stroke Detection ED
HIGH
Chest X-Ray Triage Radiology
MED
Pathology Slide Analysis Pathology
LOW
Cohort 7 Genomics Research
IDLE
Retrospective NLP Research
Active Workloads
Name
Priority
Device
Progress
Status
Stroke Detection v3.2
Viz.ai
Critical
A100 #2
Running
Path Slide QC
Paige.ai
High
A100 #1 / MIG 3g
Running
Chest X-Ray Triage
Aidoc
High
A100 #3
Running
NLP Discharge Notes
Epic + internal
Med
A100 #1 / MIG 2g
Running
Cohort 7 Genomics
Internal
Batch
V100 #3
Queued
Alerts & Events
Enclave attestation passed
GPU #4 memory pressure 91%
New workload dispatched
HIPAA audit check passed
MIG repartition
All Workloads
31 total · 24 running · 7 queued
Workload
Vendor
Dept
Priority
Device
Runtime
Status
Stroke Detection v3.2
Viz.ai
Emergency
Critical
A100 #2
8s
Running
Path Slide QC
Paige.ai
Pathology
High
A100 #1 MIG
6h 02m
Running
Chest X-Ray Triage
Aidoc
Radiology
High
A100 #3
12s
Running
NLP Discharge Notes
Internal
HIM
Med
A100 #1 MIG
2m 15s
Running
Mammography AI
iCAD
Radiology
High
A100 #3 MIG
6s
Running
Cohort 7 Genomics
Internal
Research
Batch
V100 #3
—
Queued
ICU Deterioration
Dascena
ICU
Critical
A100 #4
4d 02h
Running
Retrospective NLP
Internal
Research
Batch
—
paused
Preempted
Liver Segmentation
HeartFlow
Radiology
Med
V100 #1
18m
Running
Clinical Trial Matching
Tempus
Oncology
Med
V100 #2
4h 12m
Running
A100 #1NVIDIA A100 80GB
Healthy
MIG 3g.20gb91%
MIG 2g.10gb55%
A100 #2NVIDIA A100 80GB
Healthy
Full device72%
A100 #3NVIDIA A100 80GB
Healthy
MIG 3g.20gb88%
MIG 2g.10gb40%
A100 #4NVIDIA A100 80GB
Memory 91%
Full device95%
V100 #1NVIDIA V100 32GB
Healthy
Full device64%
V100 #2NVIDIA V100 32GB
Standby
Idle8%
Storage
NVMe cache38.2 / 48 TB
SAN (DICOM)152 / 220 TB
Audit log archive1.8 / 10 TB
Network
DICOM endpointpacs.sm.internal:11112
Egress (outbound)0 bytes
Internal throughput12.4 Gbps
Attestation heartbeat60s interval
Emergency Reservation Active
No active emergency
Reserved GPUs
3 × A100
Headroom
24 GB VRAM
Jobs preempted
2 batch · 1 research
Runtime
00:00
Scheduling Policy
How Clique routes workloads across compute
1
Clinical
Real-time patient care (stroke, ICU, ED) · preempts all
2
High
Radiology, pathology triage · SLA < 60s
3
Medium
NLP, cohort analytics · best-effort
4
Batch
Research, retrospective · runs off-hours
Department Quotas
Fair-share allocation across departments
Department
Quota
Used
Allocation
Emergency
25%
22%
Radiology
30%
28%
Pathology
15%
12%
ICU
10%
9%
Research
15%
4%
Admin / HIM
5%
3%
HIPAA
Compliant
SOC 2 Type II
Compliant
Enclaves Attested
6 / 6
Compliance Checks
Continuous verification against HIPAA, SOC 2, and internal policy
Check
Framework
Last Run
Interval
Result
Data residency (no egress)
HIPAA §164.312
2 min ago
60s
Pass
Encryption at rest (AES-256)
SOC 2 CC6.1
5 min ago
5m
Pass
Encryption in transit (TLS 1.3)
SOC 2 CC6.7
1 min ago
1m
Pass
Enclave attestation
TEE
1 min ago
60s
Pass
Access control audit (RBAC)
HIPAA §164.308
12 min ago
15m
Pass
Audit log integrity (hash chain)
HIPAA §164.312(b)
3 min ago
5m
Pass
Vendor BAA coverage
HIPAA §164.504
1h ago
1h
Pass
Patch compliance
SOC 2 CC7.1
4h ago
6h
Pass
Audit Log
Immutable, hash-chained · 7-year retention
Confidential Compute Enclaves
Hardware-attested execution environments
Enclave
Vendor
Device
Attestation
Last Verified
Status
viz-ai-stroke
NVIDIA H100 CC
A100 #2
SHA-384:a7b2...e491
38s ago
Verified
paige-path
Intel TDX
A100 #1
SHA-384:c9f1...22bd
52s ago
Verified
aidoc-cxr
NVIDIA H100 CC
A100 #3
SHA-384:3e8d...7fa0
41s ago
Verified
dascena-icu
AMD SEV-SNP
A100 #4
SHA-384:1b02...9c6e
29s ago
Verified
tempus-trial
Intel TDX
V100 #2
SHA-384:8a4c...0d13
47s ago
Verified
internal-nlp
Intel TDX
A100 #1
SHA-384:5f9e...3b88
33s ago
Verified
V
Viz.aiStroke detection, PE triage
Active · 2 models
P
Paige.aiPathology QC, prostate grading
Active · 3 models
A
AidocChest X-ray triage, C-spine
Active · 4 models
D
DascenaICU deterioration, sepsis
Active · 2 models
T
TempusClinical trial matching
Active · 1 model
I
iCADMammography AI
Active · 1 model
H
HeartFlowLiver, cardiac segmentation
Active · 2 models
Add vendor
Deploy into sealed enclave
Train a model
Drag-and-drop a labeled dataset. We'll handle the rest — architecture, batch size, validation, audit trail.
1
Drop your dataset
DICOM, NIfTI, CSV, or JSONL — up to 50 GB. Stays on-prem.
2
Pick a task
We'll auto-select the optimal pipeline based on your data and hardware.
Run inference
One-off inference against a deployed model. Results land in your EHR within seconds.
Stroke triage
Viz.ai LVO detection · v3.2.1 · attested
P95 latency 8.1s
Chest X-ray triage
Aidoc CXR · v2.4 · attested
P95 latency 12.4s
Mammography
iCAD ProFound · v4.1 · attested
P95 latency 6.2s
Discharge note NLP
Clinical-BERT · v1.7 · attested
P95 latency 2m 15s
Platform Settings
Scheduling policy
Priority order for workload routing
GPU partitioning (MIG)
Multi-instance GPU on A100 #1, #3
DICOM endpoint
PACS integration address
Attestation interval
How often enclaves are re-verified
Audit retention
HIPAA-required log retention
Egress policy — DENY ALL
Block outbound network from enclaves
Auto-preemption of batch workloads
Let clinical jobs interrupt research
Identity provider
SSO for admin console
Time zone
For scheduling and reporting