🚀
Lightweight by Design
Bring queueing and scheduling to a single machine without deploying a full cluster stack.
gflowSingle-Node Job Scheduler
Run, queue, inspect, and control GPU or CPU jobs on one Linux machine without deploying a cluster
Skip to content
📋
Flexible Submission
Submit scripts or commands with GPU requests, priorities, time limits, Conda environments, and arrays.
🔗
Dependencies and Arrays
Chain jobs together, fan out workloads, and build small research pipelines with familiar CLI commands.
📊
Queue Visibility
Inspect active or completed jobs, filter by user or project, and switch between table, tree, JSON, CSV, or YAML output.
🖥️
tmux-backed Execution
Every job runs in its own tmux session so you can attach, follow logs, and recover long-running work.
🎛️
GPU-aware Control
Restrict visible GPUs, manage shared scheduling with VRAM limits, and apply runtime controls without editing scripts.