Get Started

One execution domain. Many Garages.

Circuit coordinates available compute across multiple Garages while applications continue addressing services by logical identity.

COORDINATOR
Circuit
Garage A
6/8 lanes
us-east
Garage B
2/8 lanes
eu-west
Garage C
9/16 lanes
ap-south
Multi-Garage
Unified Domain
Cluster multiple compute nodes into one logical fabric
Dynamic
Capacity Placement
Routes to idle lanes and local artifact caches
Transparent
Logical Ingress
Clients address pit://service regardless of machine node
DISTRIBUTED COMPUTING

Smart placement without cluster overhead

Circuit turns heterogeneous machines across data centers and edge nodes into a unified execution pool.

01Scheduling

Capacity-aware Placement

Use available lanes, queue pressure and artifact locality to choose execution capacity in real-time.

02Locality

Artifact Locality

Prefer already prepared or locally cached Components when useful to minimize cross-network transfers.

03Routing

Logical Across Machines

Applications keep using pit://service regardless of which Garage executes the work. Machine boundaries stay hidden.

CIRCUIT CLUSTER MANAGEMENT

Inspect and join Garages into the Circuit

Coordinate compute hosts with simple, declarative CLI commands.

bash
# 1. Join a local or edge Garage into an active Circuit
pit circuit join --garage edge-a --token $CIRCUIT_TOKEN

# 2. Inspect active Garages, total lanes, and lane availability
pit circuit status

# 3. Route workload explicitly to optimal Garage
pit call pit://service-a/eval --placement auto
The PitFast Lifecycle

How it connects to the rest of PitFast

Every subsystem handles one distinct responsibility in the demand-driven execution chain.

DISTRIBUTED EXECUTION

Turn multiple machines into one execution domain.

Scale compute capacity across machines without managing Kubernetes nodes or complex networking meshes.