Top Builders

Explore the top contributors showcasing the highest number of app submissions within our community.

AMD Developer Cloud

AMD Developer Cloud is a cloud-based GPU platform that gives developers on-demand access to AMD Instinct accelerators. It is designed for AI researchers, engineers, and builders who need high-memory GPU compute for training, fine-tuning, and inference without managing physical hardware. Members of the AMD AI Developer Program receive $100 in credits to start building immediately.

General
AuthorAMD
TypeCloud GPU Platform
AccessAMD AI Developer Program
DocumentationAMD Developer Cloud Overview
HardwareAMD Instinct MI300X (192GB HBM3)
Pricing$100 credits for Developer Program members; pay-as-you-go available

Start building with AMD Developer Cloud

AMD Developer Cloud lets you access AMD Instinct MI300X GPUs through a simple cloud interface, so you can focus entirely on building rather than configuring infrastructure. The MI300X features 192GB of HBM3 memory, making it practical for running 70B+ parameter models on a single instance without model parallelism. Sign up for the AMD AI Developer Program to claim your credits and start running workloads today. Explore what the community has already built on AMD at AMD Use Cases and Applications.

AMD Developer Cloud Tutorials

Getting Started

Key Use Cases

Fine-tuning LLMs Use AMD Instinct MI300X instances to fine-tune open-source models such as Llama, DeepSeek, Mistral, and Qwen using PyTorch and ROCm. Hugging Face Optimum-AMD provides optimized training pipelines for AMD hardware.

Large model inference The MI300X's 192GB HBM3 memory capacity supports running very large models on a single GPU, reducing the need for multi-GPU serving setups.

Benchmarking and prototyping Test AI workloads on AMD hardware before moving to on-premises infrastructure. The pay-as-you-go pricing keeps experimentation costs low.

Hackathon development During AMD-sponsored hackathons on lablab.ai, participants receive cloud credits to access AMD GPUs directly through AMD Developer Cloud. Explore upcoming AI hackathons to find events using AMD infrastructure.

Edit on GitHub

amd AMD Developer Cloud AI technology Hackathon projects

Discover innovative solutions crafted with amd AMD Developer Cloud AI technology, developed by our community members during our engaging hackathons.

GarudaLinux

GarudaLinux

Garuda Linux is a rolling-release Linux distribution built on top of Arch Linux, designed to be more user-friendly and visually polished than vanilla Arch while retaining access to the massive Arch User Repository (AUR) for software installation. It ships with a customized KDE Plasma desktop by default (though other desktop environments like GNOME, Xfce, i3, Sway, and others are also available as separate editions), featuring a distinctive, futuristic aesthetic with translucent windows, custom themes, and vibrant color schemes.Key technical features include: BTRFS filesystem by default, paired with Snapper for automatic system snapshots — meaning if an update breaks your system, you can boot into a previous working snapshot straight from the bootloader (via Garuda's "Assistant" tool).

Simple Request Router

Simple Request Router

This project delivers an intelligent routing agent engineered to balance strict enterprise accuracy requirements against cloud API token consumption. Built to comply with an evaluation environment restricted to a 4 GB RAM and 2 vCPU footprint, the system relies on an intentional two-step isolation architecture that separates intent classification from task execution. At the core of the local optimization strategy is an ultra-lightweight, 4-bit quantized Gemma 4 (Effective 2B) parameter model running locally via llama-cpp-python. When an evaluation payload enters the pipeline, the local model acts strictly as a zero-token-cost binary classification gate. It evaluates incoming prompts across eight distinct capability domains. Direct, low-complexity natural language tasks - such as sentiment analysis, named entity recognition, basic definitions, and text summarization - are processed and answered entirely in-house, incurring a total external token cost of zero. Conversely, when the local gate detects complex algorithmic patterns, multi-step mathematical reasoning, code debugging specs, or deductive logic puzzles, it flags the task as COMPLEX and routes it to the premium Fireworks AI cloud infrastructure. By isolating the routing logic and utilizing defensive system constraints like strict stop-token structures and max-token caps, this architecture minimizes proxy token overhead while ensuring maximum accuracy for more demanding requests.

Simple Video captioner

Simple Video captioner

A two-stage video captioning agent, built reliability-first. Stage one samples keyframes from each clip and sends them to a vision-language model (Kimi K2.6 on Fireworks), which returns a dense, factual description — this is what the accuracy score is made of. Stage two turns that description into the four requested styles as a single JSON object — this is what the style score is made of. Splitting them means a caption can only be as accurate as what was actually seen, and the pressure to be funny never contaminates the grounding. Everything underneath is about not failing. A complete, valid results.json is written before the first network call, then each clip is atomically replaced as it finishes, so a crash or a hang still leaves a scoreable file behind. The agent enforces its own 540-second deadline and exits cleanly rather than waiting to be killed. Each stage walks a fallback ladder — API, then local model, then template — so no single failure is fatal. In testing, a real local-model outage still produced twelve valid captions with 292 seconds to spare. The local model is the part I most wanted to build. I distilled the teacher's style behaviour into Gemma 4 with LoRA on teacher-labelled examples, merged the adapter, and quantized it to a Q4_K_M GGUF — 5.3 GB, 2.86x smaller than bf16 — served by llama-server on CPU inside the container, no GPU required. It styles a clip in about 30 seconds on two cores. It ships as the fallback rather than the primary, because I measured instead of assuming. A blind, order-swapped LLM judge (n=19, minimum detectable effect +/-0.44) found the LoRA bought style (+0.61) but cost accuracy (-0.29), and the teacher still beat the student 10-3 on swap-surviving verdicts. The same judge found Q4_K_M statistically indistinguishable from bf16 — 4-bit quantization was free. So the distilled model earns its place by surviving an outage, not by going first.

SOAP Copilot: AI Clinical Scribe on AMD

SOAP Copilot: AI Clinical Scribe on AMD

Physicians spend roughly two hours a day on clinical documentation, time pulled directly away from patient care, and a well-documented driver of burnout that costs the US healthcare system an estimated $8.3B a year in lost productivity. SOAP Copilot is a multi-agent AI system that turns a raw doctor-patient conversation into complete, structured clinical documentation in under 30 seconds. Three specialized agents, all built on Llama 3.3 70B Instruct, handle the pipeline in sequence: a SOAP Generator produces a structured Subjective/Objective/Assessment/Plan note, an ICD Coder extracts diagnosis codes with confidence scores, and a Summary Writer rewrites the note as a plain-language patient summary. The entire stack runs on open-source models, so no PHI ever needs to leave the user's own infrastructure. The project was originally prototyped and LoRA fine-tuned on a single AMD MI300X (192GB HBM3) via AMD Developer Cloud, running vLLM on ROCm 7.2. That memory headroom is what makes serving a 70B-parameter model in BF16 precision practical on one chip, something no consumer GPU and few cloud GPUs can do. For this submission, production inference is served through Fireworks AI, which itself runs on AMD Instinct hardware, so the project demonstrates the AMD AI stack end to end: self-managed fine-tuning through to hosted inference. Roadmap items include a distilled specialist model for lower-cost deployment, retrieval-augmented generation over clinical guidelines using the Qdrant vector database already wired into the architecture, and an on-premise, HIPAA-compliant deployment path for health systems that cannot send patient data to any external API.

Thymus

Thymus

Thymus is a lightweight hybrid token-efficient router designed to maximize accuracy while minimizing token costs in multi‑task LLM pipelines. It dynamically routes user queries across local and remote models on LLM providers. The general architecture of the router is as follows. 1. Classify: a fast regex/keyword pass handles the obvious cases for free; anything ambiguous falls back to the local model for classification. 2. Infer locally: the local model (Qwen2.5-3B-Instruct, GGUF, CPU) answers the query and self-reports a routing preference and confidence. 3. Verify: a task-aware verifier sanity-checks the answer (JSON structure, code execution, entity coverage, self-consistency, depending on task type). 4. Decide: LOCAL (free, done), SHEPHERD (local answer used as context, one cheap cloud pass to double check), or RESET (send straight to Fireworks).

CrosslaneAI

CrosslaneAI

CrosslaneAI is an agentic migration platform that autonomously ports CUDA codebases to ROCm/HIP, so teams can run their GPU workloads on AMD hardware without rewriting a single kernel by hand. For over fifteen years, the entire GPU computing world has been written in CUDA. That accumulated codebase — not the silicon itself — is the real reason teams stay locked to a single vendor. AMD's MI300X is genuinely competitive today, offering 192GB of HBM3 and far better price-per-hour economics, yet most organizations can't make the switch. The migration is manual, slow, error-prone, and expensive enough that it's typically handed to specialist consultants charging weeks of engineering time. PortForge productizes that migration and collapses it from weeks into minutes.