Open Source Workflow Orchestration for Heterogenous Computing
Make your experiments more manageable, scalable and reproducible.
Manage, deploy & scale workloads across the world’s most advanced computing hardware
import covalent as ct @ct.electron def add_space(name): return name + ' ' @ct.electron def join(a): return ''.join(a) @ct.lattice def workflow(name): name_space=add_space(name) result = join('Hello', name_space) return result + ' !!' id = ct.dispatch(workflow)(name='Shore') ct.get_result(id, wait=True)
Covalent is purpose-built for heterogeneous workflows; hardware, software and clouds
Covalent makes workflows scalable
- Serverless HPC: don’t waste resources while waiting in queue
- Code locally and scale anywhere from on prem HPC to the cloud
- Native parallelization across diverse hardware types
Covalent makes workflows manageable
- Visualize and monitor workflows in Covalent’s browser-based UI
- Manage experiments across heterogeneous hardware and software
- Write in Python, not YAML
Covalent makes workflows reproducible
- Version your experiments just as you would version your code.
- Easily store and retrieve past runs and share with colleagues
- Built-in checkpointing capabilities to prevent re-runs
Who is Covalent for?
Computational scientists running large-scale experiments
ML and Data Scientists dealing with heterogeneous computing tasks
Enterprises & Teams
Enterprises and research labs working across hardware types and clouds