Hey everyone! A few of us have been building a tool over the past year to simplify running compute-heavy workloads in the cloud. It started as something we made for internal research teams who didn’t want to spend time wiring up infra every time they ran a job, but we’re curious if it’s useful to a wider developer/DevOps audience.
The problem we kept running into...
Spinning up cloud resources for workloads: ML training, data pipelines, simulations, etc. usually requires a decent amount of DevOps plumbing...
-provisioning instances
-setting up environments
-handling IAM + networking
-writing Terraform
-scaling rules
-cleanup afterward
A lot of folks told us they just wanted to run their code and not think about infra for every job.
What we built: We created Adviser, which lets you take any command you normally run and just prepend:
adviser run
Examples:
adviser run python train.py
adviser run Rscript analysis.R
adviser run ./simulate
Adviser handles provisioning, environment setup, cost/performance optimization, scaling, and teardown automatically. can be Python, R, C++, Rust, shell scripts - anything.
Who this is for
-ML engineers, Data scientists, devops and platform engineers supporting heavy workloads, researchers running simulations or pipelines
Why we’re posting here:
We’re opening up a self-serve version and genuinely want feedback from folks who run cloud workloads professionally. If this seems redundant, missing obvious features, or solves a real headache...all perspectives are helpful.
To make testing easier, we’re giving every new user $100 of free compute.
Links
Docs: https://github.com/adviserlabs/docs
Slack: https://www.adviser.sh/
Demo: https://www.youtube.com/shorts/F9feaOr3TbA
Happy to answer any questions or hear thoughts on what would make this more useful.
Top comments (0)