Building PrivateDAO Agent Exchange on AMD — Hello AMD Developer Community

Hi AMD Developer Community,

I’m Fahd Kotb, Founder of PrivateDAO. This is my first post here, so I wanted to introduce what we’re building and why I joined the AMD developer ecosystem.

PrivateDAO is building privacy-preserving infrastructure for organizations and AI agents — combining private execution, policy-controlled workflows and verifiable outcomes.

One of our core products is PrivateDAO Agent Exchange: https://agents.privatedao.org⁠

It is a production platform where developers, organizations and AI agents can discover and execute agent services through APIs and MCP, with policy controls and verifiable execution receipts.

I’m currently exploring how we can bring AMD GPUs, ROCm and AMD AI infrastructure into the Agent Exchange for AI inference, agent workloads and GPU-accelerated services.

A little background on our journey:

• Member of Superteam Poland • 1st Place — Superteam Poland Rebuild Hackathon • 3rd Place — Superteam UAE Frontier Hackathon • Currently participating in the Crypto World’s Fair • Member of the GitHub Developer Program • PrivateDAO is a registered IBM Partner Plus member and has joined the IBM watsonx Orchestrate Agent Connect Partner Program • PrivateDAO Agent Exchange is available on GitHub Marketplace

I built PrivateDAO founder-led, and my goal now is simple: connect what we have already built with the best infrastructure available and keep shipping.

I’d especially love to connect with AMD developers working around ROCm, AMD Instinct, AI inference and agent infrastructure.

Looking forward to building with the AMD community.

Fahd Kotb

Founder & Owner — PrivateDAO

Hello, thanks alot for sharing. Since your inference side is already built around provider paths, the cleanest way in is an OpenAI-compatible one: vLLM on Instinct, and llama.cpp or AMD’s Lemonade Server on Radeon and Ryzen AI, all expose that API, so one adapter covers the whole range. For Instinct without buying hardware, AMD Developer Cloud rents MI300X instances: https://devcloud.amd.com If you’re not already in the AMD AI Developer Program, it’s free to join and comes with cloud credits, and members who share projects publicly can apply for more: AMD AI Developer Program A member here also built an MCP server that manages those instances and scaffolds vLLM, which fits the Exchange nicely: GitHub - xbill9/amd-gputools: Python + shell tooling for AMD Instinct MI300 GPUs on DigitalOcean, with an MCP server for droplet lifecycle and remote ROCm access · GitHub

1 Like

Hi @fahd_kotb, welcome to the community! Combining privacy-preserving execution with AI agent infrastructure sounds like a really compelling use case. Looking forward to seeing how PrivateDAO leverages ROCm and AMD hardware!

1 Like

Thank you Hakob and Dorothy

This is exactly the kind of technical feedback I was hoping for

PrivateDAO Agent Exchange already has its agent infrastructure and MCP layer in place, so an OpenAI-compatible AMD inference adapter looks like a practical next step

I’ll explore vLLM on AMD Instinct and Lemonade for Radeon and Ryzen AI, starting with a small reproducible integration

I’d also love to explore AMD Developer Cloud credits and collaborate with developers working on ROCm and MCP infrastructure

My goal is to make AMD-powered inference accessible to organizations and agents through PrivateDAO’s existing workflows

Appreciate the warm welcome and

Thank you Hakob and Dorothy

This is exactly the kind of technical feedback I was hoping for

PrivateDAO Agent Exchange already has its agent infrastructure and MCP layer in place, so an OpenAI-compatible AMD inference adapter looks like a practical next step

I’ll explore vLLM on AMD Instinct and Lemonade for Radeon and Ryzen AI, starting with a small reproducible integration

I’d also love to explore AMD Developer Cloud credits and collaborate with developers working on ROCm and MCP infrastructure

My goal is to make AMD-powered inference accessible to organizations and agents through PrivateDAO’s existing workflows

Appreciate the warm welcome and

Thank you Hakob and Dorothy

This is exactly the kind of technical feedback I was hoping for

PrivateDAO Agent Exchange already has its agent infrastructure and MCP layer in place, so an OpenAI-compatible AMD inference adapter looks like a practical next step

I’ll explore vLLM on AMD Instinct and Lemonade for Radeon and Ryzen AI, starting with a small reproducible integration

I’d also love to explore AMD Developer Cloud credits and collaborate with developers working on ROCm and MCP infrastructure

My goal is to make AMD-powered inference accessible to organizations and agents through PrivateDAO’s existing workflows

Appreciate the warm welcome and the guidancethe guidancethe guidance

1 Like