Mernit

8 ms·

1.
Show HN: Agent Sandboxes with GPUs from $0.20/hr(tama.computer)
5 points by Mernit 13d ago|1 comments
2.
by Mernit 1mo ago
Replicate was acquired by Cloudflare last fall [1] If you're looking for an alternative, I work at beam.cloud which is another serverless GPU cloud with fast cold starts. [1] https://www.cloudflare.com/press/press-
3.
What Is a Container, Really? Five Years of GPU Infrastructure(beam.cloud)
1 points by Mernit 2mo ago|0 comments
4.
Your Company is a Filesystem(twitter.com)
2 points by Mernit 7mo ago|0 comments
5.
Before the Super Bowl, Seahawks kicker Jason Myers was so calm he felt nervous(nytimes.com)
1 points by Mernit 7mo ago|0 comments
6.
Why Most of America Is Terrible at Making Biscuits (2018)(theatlantic.com)
5 points by Mernit 8mo ago|1 comments
7.
by Mernit 1y ago
We initially wrote this for a less technical audience (where we spelled out MCP), then edited it to post here - it's not AI, it's just bad editing from my part. Fixed now.
8.
by Mernit 1y ago
The only Beam-specific part are the sandboxes, but those can easily be swapped out for the vendor of your choice. The architecture we described isn't exclusive to our product. Beam is fully OSS BTW: https://github.com/b
9.
by Mernit 1y ago
what is missing for you?
10.
Building Proactive AI Agents(substack.com)
2 points by Mernit 1y ago|0 comments
11.
The Architecture Behind Lovable and Bolt(beam.cloud)
73 points by Mernit 1y ago|29 comments
12.
by Mernit 1y ago
beam.cloud (YC W22) | Product Engineer, SWE, SRE, Infrastructure, DevOps, Networking | NYC | Onsite Beam is an ultrafast AI inference platform. We built a serverless runtime that launches GPU-backed containers in less than 1 second and quic
13.
by Mernit 1y ago
You should look into beam.cloud (I'm the founder, but it's pretty great) It lets you quickly run long-running jobs on the cloud by adding a simple decorator to your Python code: from beam import function # Some long trainin
14.
by Mernit 2y ago
You should checkout beam.cloud (I'm the founder). It's a modern FaaS platform for Python, with support for REST endpoints, task queues, scheduled jobs, and GPU support.
15.
Show HN: Run GPU Workloads Across Multiple Clouds(github.com)
4 points by Mernit 2y ago|1 comments
16.
by Mernit 2y ago
The major clouds don't support serverless GPU because the architecture is fundamentally different from running CPU workloads. For Lambda specifically, there's no way of running multiple customer workloads on a single GPU with Fire
17.
Building a Modern Serverless Cloud for Bioinformatics(beam.cloud)
1 points by Mernit 2y ago|0 comments
18.
by Mernit 2y ago
beam.cloud | Founding Software Engineer, Infrastructure | Full-time | REMOTE | New York, NY USA Beam is building a cloud runtime for running remote containers on GPUs. We’re used by thousands of developers for powering their generative AI a
19.
by Mernit 2y ago
Yes we are hiring internationally
20.
by Mernit 2y ago
beam.cloud | Founding Software Engineer, Infrastructure | Full-time | REMOTE | New York, NY USA Beam is building a cloud runtime for running remote containers on GPUs. We’re used by thousands of developers for powering their generative AI a
21.
by Mernit 2y ago
There are a number of good options here. The different axes are cost of GPUs, performance, and ease of use / developer experience. You might consider beam.cloud (I'm one of the founders), which is oriented strongly on the performa
22.
Beam.cloud (YC W22) is hiring engineers to build a GPU container runtime(ycombinator.com)
1 points by Mernit 2y ago
23.
by Mernit 3y ago
You should checkout https://beam.cloud (I'm the founder), it'll give you access to plenty of cloud GPU resources for training or inference. Right now it's pretty hard to get GPU quota on AWS/GCP, so hopefully
24.
by Mernit 3y ago
Cloudflare AI and Replicate are great for running off-the-shelf models, but anything custom is going to incur a 10+ minute cold start. For running custom fine-tuned models on serverless, you could look into https://beam.cloud wh
25.
by Mernit 3y ago
Serverless only works if the cold boot is fast. For context, my company runs a serverless cloud GPU product called https://beam.cloud , which we've optimized for fast cold start. We see Whisper in production cold start in un
26.
by Mernit 3y ago
Sharing GPUs only really makes sense for GPUs that are large enough to share. MIGs can work for 80Gi A100s but won't work with smaller cards like T4s. It also adds latency to the GPU operations. Unfortunately there's not yet a sil
27.
by Mernit 3y ago
What are you using for K8s autoscaling? We initially tried a few standard K8s scaling mechanisms and found that they didn't work well for GPU workloads. For example, if we were serving a low-RAM Huggingface model on GPU, it wouldn&#x27
28.
Run CodeLlama on a Serverless GPU(github.com)
3 points by Mernit 3y ago|0 comments
29.
by Mernit 3y ago
You could easily host your model on https://beam.cloud (I'm a founder). You just add a decorator to your existing Python code: from beam import App, Runtime app = App(name="gpu-app", runtime=Runtime(gpu
30.
by Mernit 3y ago
You could try https://beam.cloud (disclaimer, I’m the co-founder) You can run popular models on remote GPUs from your laptop with two commands. For example: $ beam create-app stable-diffusion-gpu $ beam start app.py

More ›