Advancing AI innovation together. Built with devs, for devs. Supported through an open ecosystem. Powered by AMD. #TogetherWeAdvance
- Tweets 2,010
- Following 115
- Followers 58,731
- Likes 1,367
Open collaboration is moving GPU kernel development forward.
AMD and contributors from @Caltech, @Stanford, @togethercompute and the broader developer community teamed up to bring HipKittens support to AMD Instinct MI455X GPUs. The collaboration produced peak-performance GEMM kernels for the Helios GPU architecture using HipKittens.
The walkthrough starts with a baseline kernel and progresses through optimizations including asynchronous memory transfers, Tensor Data Movers (TDM) and workgroup-cluster multicast. Developers can follow the code and see how each technique shapes kernel performance.
Explore the technical breakdown and code: rocm.blogs.amd.com/software-…
7
42
2,303
AI at AMD retweeted
🎙️ We chat with @AnushElangovan, VP of Software at @AMD, about ROCm's open-source unified toolchain for GPUs, how agentic AI is drastically lowering the barrier to entry for low-level hardware programming, and the rapid convergence of software and hardware development timelines.
stackoverflow.blog/2026/09/2…
2
6
35
5,952
How fast can an AI agent work as its context grows? @TileRT_AI reached 469 tokens per second with GLM-5.3 on 8 AMD Instinct™ MI355X GPUs
Explore how TileRT and AMD are advancing inference for long-running agent workloads: amd.com/en/developer/resourc…
2 ways to win on OpenRouter, serving GLM5.3/MI355X:
1. Speed: TileRT hits 469 tps/user on AgentX, ~45% higher interactivity v. GB300/FP4: amd.com/en/developer/resourc…
2. Profits: $11.18/GPU/hr at 67 tps (SemiAnalysis AgentX, 60% utilization) => $11.18*24*265*8 = $560k profit/node/year
1
3
21
2,163
The reviews are in…
The first participants to take the ROCm Associate Certification put it to the test, building their knowledge and applying new skills.
The verdict: worth it.
Get certified online: developer.amd.com/rocm-certi…
2
9
30
2,078
AI at AMD retweeted
we got early access and hipkittens is now running on amd helios mi455! here we share an educational gemm ladder to demonstrate how to use the new helios features.
overall, we found that the patterns identified in hipkittens for writing performant mi350/355 kernels translate well to helios making the forward port quite smooth!
Made with AI
2
24
154
13,681
Have you started your ROCm certification yet?
The AMD ROCm Certified Program has officially launched on AMD AI Academy! This self-paced certification program is designed to help developers, data scientists, and technical teams build and validate practical ROCm™ skills for AI and HPC workloads on AMD GPUs.
Build your ROCm expertise and validate your skills today! developer.amd.com/rocm-certi…
1
1
17
2,206
10 things to know before joining the AMD AI Developer Program 👇
1. It’s free. Anyone can join, whether you are a student starting out or a seasoned ML engineer.
2. You get $100 in cloud credits on day one. Enough to prototype, benchmark, and get moving on AMD tech.
3. You don’t need any hardware to start. Build immediately with cloud access.
4. There’s a private Discord with a community of engineers who will provide technical discussion and help.
5. You get early invites to dev events and workshops. Before they’re public. Before they fill up.
6. Monthly Ask the Expert hours with open-source partners. Ask questions. Go deep. Learn the tricks of the trade.
7. Your projects can get featured on official AMD Developer channels and at events.
8. You’re automatically entered into monthly hardware raffles. No forms. No extra steps.
9. You get a free month of @DeepLearningAI Pro. High quality courses for you to expand your skills.
10. The more you participate, the more you unlock. More access, more visibility, more opportunities.
If you’re building, training, or optimizing AI models, we’d love to have you join us!
23
83
713
37,551
AI at AMD retweeted
Four talks, one hands-on workshop, Oct 8 in SF.
@AIatAMD, @cerebras, @moonmathai, and @RedHat on what it actually takes to run inference at scale with @vllm_project and @llm_d.
Bring a laptop. Agenda below. #SFTechWeek
RSVP: partiful.com/e/63yfXybwQqlyQ…
ALT vLLM & llm-d inference meetup during the San Francisco Tech Week, on October 8, 2026.
2
6
30
2,018
14,000 models. One optimization pass.
See how AMD ROCm Hyperloom uses LLMs to explore search spaces, identify hotspots, and help unlock workload performance.
@AnushElangovan explains on @chain_ofthought podcast with @ConorBronsdon: piped.video/watch?v=xMC7d_h5…
4
6
24
2,343
Going to #ROSCon2026?
Join AMD for Train, Simulate, Deploy: Agentic AI from Cloud to Robot, a workshop exploring the tools and workflows behind building and deploying agentic AI for robotics.
📍 ROSCon 2026, Toronto
📅 September 22–24
You can also stop by the AMD booth throughout the conference to connect with the team and learn more about open development for robotics.
Can’t make it to Toronto? AMD is supporting the ROSCon livestream and video archive so you can follow along remotely.
Learn more: roscon.ros.org/
ALT The image is a promotional graphic for "ROSCon Global 2026" in Toronto, from September 22nd to 24th. It features cartoon characters in hockey gear, including a turtle holding a hockey puck with the ROS logo. The text mentions AMD as the Platinum sponsor and includes the tagline “together we advance.” The background features a silhouette of a cityscape.
2
4
11
2,376
IFA Opening Keynote, presented by Jack Huynh nitter.net/i/broadcasts/1yxBePboN…
1
4
22
2,244
The Case for Graph-Based AI Agents
AI development is evolving from prompt engineering → context engineering → agent loops → graph engineering.
The key shift: instead of asking one agent to reason over everything, graph engineering gives different agents focused responsibilities and connects them through a deliberate workflow.
In this tutorial,
- Three specialist agents independently analyze logs, metrics, and traces
- A coordinator correlates the findings, without doing the analysis itself
The result? It identifies a CPU-contention fault that the noisiest log source completely hides.
Built with OpenClaw and vLLM, running a local Qwen3.6-35B-A3B model on a single AMD Instinct™ MI300X GPU.
The takeaway: AI reliability isn’t just about better models or more context. It’s also about designing the right structure for how agents’ reason.
rocm.docs.amd.com/projects/a…
4
5
39
2,314
The AMD ROCm Certified Program has officially launched on AMD AI Academy! This self-paced certification program is designed to help developers, data scientists, and technical teams build and validate practical ROCm skills for AI and HPC workloads on AMD GPUs.
Build your ROCm expertise and validate your skills today! developer.amd.com/rocm-certi…
3
17
94
4,462
Three tracks. Real AMD hardware. 20,000+ builders.
The AMD Developer Hackathon: ACT II is a wrap - and these are the ones who took the podium ↓
lablab.ai/ai-hackathons/amd-….
4
3
15
1,751
Ready to build what’s next in AI? Join the AMD AI Developer Program and connect with thousands of developers like Mriganka Sarkar who are turning bold ideas into real-world AI solutions on AMD technology.
Join today: developer.amd.com/ai-develop…
ALT A testimonial from Mriganka Sarkar, part of the AMD AI Developer Program. It describes gaining hands-on experience with open source models on GPUs, learning through building, and creating a pricing software using AMD models. The AMD AI Developer Program logo is at the bottom.
2
12
2,159
Qwen 3.8 27B is already agentic lore.
So we ran two Qwen 3.8 27B models in one local build.
We stacked the Qwen models using the new Windows Hermes Agent desktop bot mode running on our 128 GB Ryzen AI Max+ processor based agent computer.
See how Ryzen AI Max+ architecture keeps local Qwen agent builders, building. amd.com/en/blogs/2026/run-qw…
29
56
608
45,034
Nearly 3,900 developers took on the AMD DevMaster Hackathon across Multimodal, Agentic and Physical AI. The biggest lesson? Great AI goes beyond powerful models—it solves real problems, executes reliably and connects intelligence to the real world.
Explore the winning projects and key takeaways: amd.com/en/developer/resourc…
2
17
2,420