Space data centers built on LEO constellations create an opportunity to serve Compound AI workflows close to where observations are captured. Serving such a workflow for an area of interest (AoI) requires deciding which orbital planes to provision and where to place each workflow component, while orbital motion continuously changes which satellites can serve the AoI. Provisioning must therefore be planned as a sequence of decisions over time, and these decisions are coupled because changing component placements or activating new planes incurs costs. To address this, we present Clockwork, a framework for scheduling Compound AI workflows on space data centers under end-to-end latency SLOs. Clockwork has two components: i) Castor jointly selects active planes and component placements for each constellation snapshot, while ii) Pollux connects Castor's candidate provisioning states across predicted orbital evolution to construct a low-cost schedule over time. On our experimental two-shell constellation, Castor serves 81% of instances at a 300 ms SLO, while Pollux reduces mean total scheduling cost by 56% on average and by up to 63% compared to heuristic baselines.
Figures & tables
Fig. 1: Wildfire detection use case
Fig. 2: Clockwork overview. (1) The provider offers shells of orbital planes with heterogeneous hardware; a request names an AoI, a Compound AI system workflow, and a latency SLO. (2) Per snapshot, Castor searches a layered graph of candidate satellites for a minimum-cost feasible state and retains near-minimum-cost alternatives. (3) Per cycle, Pollux selects a state sequence from the resulting pool, using future snapshots to minimize necessary reprovisionings and cost.
Parameter
Default value
Orbital substrate
Star shell
12×16 , 561 km, 87∘ ; 168 CPU, 20 small, 4 large acc.
Delta shell
12×16 , 888 km, 53∘ ; 179 CPU, 12 small, 1 large acc.
Repeat cycle
One sidereal day ( 86164 s): 15 Star, 14 Delta revolutions
CPU / small / large 1 / 4 / 10, active plane 5 units per hour
TABLE I: Default experimental configuration. Shell dimensions give planes × satellites per plane.
Output
Execution latency (ms)
Component
Model
Capacity
(MB)
CPU
Small
Large
share
accel.
accel.
Detector
YOLOv8m
0.50
2.80
90.1
96.6
9.0
Classifier
ResNet-50
0.25
0.04
18.7
24.6
2.1
LLM
Qwen2.5-0.5B
0.75
0.01
n/a
900
117
TABLE II: Compound AI workflow and execution profiles
Fig. 3: Effect of the latency SLO on snapshot provisioning. (a) Feasibility vs various SLOs. (b) and (c) Fraction of instances served at or below each provisioning cost at the 300 ms and 1500 ms SLO.
Fig. 4: (a) Mean total cost across three seeded hardware configurations, split into provisioning (solid) and reprovisioning (hatched) cost. (b) Cumulative cost during the first two hours for one seeded hardware configuration. Gray bands mark periods with no SLO feasible candidate plan.
Fig. 5: Effect of accelerator availability on placement.
Day-two SLO attainment (%)
Reprov.
Shell altitudes (km)
Ceiling
Replay
Replan
Repeat, 561/888
84.3
84.3
84.3
62
Non-repeat, 550/800
83.7
51.4 [49–53]
83.7
61
Non-repeat, 600/1000
85.8
47.1 [42–50]
85.8
59
TABLE III: Effect of orbital repeatability on the Pollux schedule. Brackets give the min and max replay result across seeds.
Model
HW
Orbital
Plans
Prov.
selection
classes
motion
ahead
cost
Compound AI serving [ 28 , 29 , 30 , 9 , 31 ]
✓
(✓)
–
–
✓
LEO placement [ 13 , 33 , 14 , 16 ]
–
(✓)
✓
(✓)
(✓)
Time-expanded chains [ 17 , 18 , 19 ]
–
–
✓
✓
✓
HyperDrive [ 11 ]
–
✓
(✓)
–
–
Clockwork
–
✓
✓
✓
✓
TABLE IV: Related work by group. ✓ all systems in the group, (✓) some, – none. Columns: model selection; hardware heterogeneity; placement feasibility with orbital motion; placements planned ahead over snapshots; provisioning cost in the objective.