My main use case for Coder is providing desktop-like environments for access into our various clusters. A quick, specific example of how I use Coder for that is through our Jupiter Hub integrations, where we frequently use VS Code to enable researchers to perform ad hoc research on their laptops while also being able to run DAGs in Argo workflows against our GPU clusters.
What is our primary use case?
What is most valuable?
The best features Coder offers include easy integration into a remote environment that is accessible from everywhere, VS Code integrations, and AI integrations. The AI integrations I mentioned include various ways people access AI, such as using Cursor and Claude with various models. We have fairly extensive licensing as we are a biologics company, so the integration with Anthropic's Life Sciences has been wonderful as well as the integration with cloud workflow.
Coder has positively impacted my organization by providing a central place for people to run their research and enabling us as administrators to focus on one aspect of researcher interactions with our clusters, so we do not have to provide much education or options for integrating their workflows into our clusters.
What needs improvement?
From my perspective as an administrator, Coder is very valuable for being able to get access to the work that people are doing so I can troubleshoot it. As a general rule, it is more than sufficient for my needs, and how Coder can be improved is really a researcher question.
For how long have I used the solution?
I have been using Coder for the last two or three years.
What do I think about the stability of the solution?
Coder is very stable and incredibly reliable.
What do I think about the scalability of the solution?
Coder's scalability is great, allowing us to allocate more resources to it with a simple scheduler change, which enables us to add as many researchers as we need.
How are customer service and support?
The customer support is excellent and superb. I would rate the customer support a 10 on a scale of 1 to 10.
Which solution did I use previously and why did I switch?
I previously used Jupiter Hub, which we ran integrated into a self-hosted environment. It was limited because many people preferred a different kind of shell environment, and they had been passing Jupiter Hub notebooks between each other, which was not conducive to a productionized environment.
How was the initial setup?
Coder helps with onboarding new users or researchers by making it easy once they are in the system. We just tell them how to log in and they have single sign-on access. It is very easy.
What was our ROI?
If anything, the ROI would be time saved, although I do not have metrics for anything else.
What's my experience with pricing, setup cost, and licensing?
I was not really involved with the pricing or the setup costs, but I know that we have an advantaged per-seat licensing because we were early into it. The payment process is easy for us, and it provides so much value in our company that I cannot see doing without it.
Which other solutions did I evaluate?
If we did evaluate other options before choosing Coder, I do not remember what they were, but there was nothing really out there that answered our questions, especially after the frustrations with Jupiter Hub.
What other advice do I have?
I think Coder is very accurate and reliable in terms of output and is extremely reliable as the base for our infrastructure and infrastructure interaction.
Coder is deployed on-premises in my organization.
Coder integrates with our existing infrastructure as the main interface for our two clusters that have GPUs—some are AMD GPUs and the other one is a NVIDIA GPU system at Oracle. Users can access either cluster through Coder's interface, whether using VS Code, Cursor integration, or whatever they choose, from their laptops wherever they are. It is the starting point for any researcher doing research, and it is very simple to get them enabled and set up.
Coder handles user authentication and access control in our organization through Okta integration, where Okta groups are managed by our IT team, allowing access only through certain groups within Okta, so that access is locked down.
Collaboration between researchers or teams is supported by Coder, as they can share notebooks. Every Coder instance that comes up also has a shared file system, a high-speed shared file system behind it, which is incredibly helpful.
Coder enables integration with third-party tools or services through a variety of plugins, with the VS Code one probably being the most used. I know that Cursor, the Cursor UI also has an integration into it, and we do allow Jupiter Hub notebooks, but I do not know if we have many people using those anymore.
Coder helps me manage resources, such as compute or storage, by limiting the size of Coder notebook spaces. We are able to give everybody a GPU or more based on the options in the templates, and the templating system is incredibly valuable for us because it allows us to provide a working environment that is sufficiently powerful but has limits so that people do not overrun the infrastructure allocated to them.
Coder's AI capabilities are very secure, as we have Okta integration so that we are locked down from the perspective of access. It allows for fine-grain controls for users, and we run it self-hosted, making it an ideal self-hosting candidate. It enables people from all over the world to have one-step integration from their laptops wherever they are in the world, and it is very effective for everyone, given our remote teams.
My advice to others looking into using Coder is to commit to the decision without hesitation. I would rate this overall experience a 9 out of 10.
Which deployment model are you using for this solution?
On-premises
If public cloud, private cloud, or hybrid cloud, which cloud provider do you use?
Other

