Filters

AI Models

Data

Redis Qdrant

Platforms

Streamlit Replit Vercel

Upcoming AI Hackathons
For Innovators & Creators

Meta LLaMA Applications

Browse applications built on Meta LLaMA technology. Explore PoC and MVP applications created by our community and discover innovative use cases for Meta LLaMA technology.

SOAP Copilot: AI Clinical Scribe on AMD

SOAP Copilot turns raw doctor-patient conversations into structured SOAP notes, ICD-10 codes, and patient-friendly summaries in seconds, using a 3-agent Llama 3.3 70B pipeline built and fine-tuned on AMD hardware.

LoneSoloWolf

AMD Developer CloudAMD ROCmLLaMA+1

SmartBandit Router

A reinforcement-learning router that learns which AI model to call for each query, cutting cost by ~93% without ever hardcoding a routing rule.

nyan

LLaMAQwen3GPT-3

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 Disappointer

HuggingFace HubLLaMAAMD Developer Cloud

Frugal Router: Cost-Aware Multi-Model AI Agent

A category-aware router for the 8 Track 1 task types: sentiment/NER/factual questions go to a bundled local model or the cheapest Fireworks tier; math scales by difficulty; code and logic puzzles route straight to frontier.

Lisan AL Gaib

AMD Developer CloudAMD ROCmDeepSeek R1+9

Cyber Sentinel AI

Cyber Sentinel AI uses multiple AI agents to investigate suspicious emails, URLs, prompt injection attacks, and browser extensions, delivering threat analysis, independent verification, security scoring, and actionable response recommendations.

DhurandharX

GemmaGrokOpenAI+1

YouTube Growth Copilot

AI-powered YouTube analytics platform that transforms real channel data into actionable growth strategies using deterministic analytics, machine learning, and Llama-powered insights.

katana

AI/ML APIGroqLLaMA

DevMemAgent: Dev Memory on AMD ROCm

A typed, decaying, graph-aware memory layer for software engineers that runs fully local on AMD ROCm / Instinct GPUs — no cloud, no data leaves the machine.

The Phalanx

AMD ROCmLangChainLlama 3.1+5

Confidence-Cascade Routing Agent

A highly token-efficient AI routing agent that solves tasks using a free local Gemma 4B model, only escalating to a remote 31B model via Fireworks AI when strict confidence thresholds fail.

RooAnu

AMD Developer CloudAntigravityGemma+1

RugPullRadar

On-chain rug-pull risk scanner for ERC-20 tokens - five auditable signals plus a plain-English verdict served by an LLM on AMD Instinct.

AMDeus Ex Machina

AMD ROCmAMD Developer CloudGPT-5+2

StormSense AI

Real-time disaster intelligence platform powered by multi-agent AI. Autonomously monitors live earthquake, flood and wildfire data from USGS, NASA FIRMS and OpenWeatherMap, assesses risk and delivers instant alerts.

StormSense AI

AgentOpsAI/ML APIAMD Developer Cloud+12

Hyper Efficient 0 token Router v2

An ultra-fast, local LLM evaluation container running Private trained model via optimized llama-cpp-python. Reconfigured to achieve sub-8 second per-task processing speeds under a strict 4GB RAM threshold

Absolut Void Development

HuggingFace HubGithub CopilotGemini 3 Flash+2

MediSense AMD — Multimodal Clinical AI Agent

Production AI healthcare platform serving 869+ real users across 9 countries — 4 ML models + LLM on AMD MI300X via Fireworks AI, bilingual Hindi/English with WhatsApp alerts.

MediSense AMD

AI/ML APIGroqLLaMA+1

PulseBoard

Autonomous multi-agent AI system that analyzes your business data and delivers a plain-English morning brief before you wake up. 5 specialist agents run in parallel on AMD Developer Cloud via Fireworks AI.

Cogent

AMD Developer CloudAMD ROCmLLaMA

Hybrid Token-Efficient Routing Agent

A CPU-only AI agent combining a local SmolLM3-3B model with selective Fireworks API escalation, aiming for zero-token accuracy across 8 task categories under a 2 vCPU / 4 GB grading environment.

AmyA

LLaMArest api

AMD Smart Router + Video Caption Agent

An optimized dual-track AI system featuring a zero-token local heuristic router to minimize API costs using Gemma & Qwen on AMD infrastructure, paired with an offline, resource-efficient 4-style video captioning pipeline running entirely in Docker.

Power SASE CR

AMD Developer CloudAMD ROCmGemma+1

MarketScout AI

MarketScout AI is an autonomous multi-agent platform that transforms startup ideas into market intelligence. Powered by Fireworks AI on AMD Cloud, it researches competitors, patents, funding, trends, and strategy to validate ideas in minutes.

Hack-Horizon

AMD Developer CloudlangflowLLaMA+4

NanoRouter

A highly token-efficient, hybrid AI agent that uses local-first processing (Qwen 3.5 0.8B) for basic tasks and smart structural routing to escalate complex logic and code to the Fireworks API.

The Overclocked

Qwen3LLaMA

AI-Quant-Coder

AI tool to predict financial markets and automatically generate Python trading scripts using open-source LLMs.

AI Quant-Coder

Mistral AILLaMAAlpaca+5

Haven

Haven is an AI mental wellness companion: empathetic chat, journaling insights, voice therapy, breathing exercises, and clinical tools like PHQ-9 and GAD-7 — powered by open-source Llama & Qwen models with layered crisis detection.

Ghost Coders

AI/ML APIAMD Developer CloudAMD ROCm+8

Fairwind

Fairwind is a hybrid token-efficient routing agent for AMD Developer Hackathon Act II Track 1. It classifies each task into one of 8 categories and answers it via the cheapest reliable tier available.

Heisenberg

Anthropic ClaudeClaude CodeLLaMA

AI Gateway: Hybrid Token Routing Agent

An intelligent AI gateway that uses LangGraph to route requests across local Gemma, GPT, Claude, and Gemini models based on cost, latency, privacy, quality, and token efficiency, with explainable routing and analytics.

zynai

AI/ML APILLaMAChatGPT+6

Cheapskate Router

A containerized AI agent that routes tasks across 8 capability categories to the cheapest sufficient Fireworks AI model, using zero-token local classification and syntax validation to minimize token spend while clearing the accuracy gate.

Crazybell

AMD Developer CloudAnthropic ClaudeAntigravity+5

ZeroFire — Zero-Token Local-First Routing Agent 02

A Track 1 agent that answers all 8 task categories with 0 Fireworks tokens: an in-container Qwen2.5-1.5B plus deterministic verification — programs are executed, code compiled and run, logic solved in code, every answer checked before it is trusted.

PizzaDiet&ColdWater Enterprise

LLaMAQwen3AMD Developer Cloud

KNOX (Knowledge-driven Natural Execution System)

Knowledge-driven Natural Execution System (KNOX): A smart assistant system that uses AI-based knowledge and natural inputs like voice and gestures to understand user requests and execute actions automatically.

ZEROONE

GemmaAI/ML APILLaMA

ZeroToken Router

A zero-token-first hybrid AI agent that routes 8 task categories between deterministic local solvers, a bundled Qwen model, and Fireworks API models. Maximizes accuracy (100% on validation sets) while reducing external token costs by over 52%.

Gitstroyers

LLaMACodexrest api

Rauto

Rauto is an intelligent, token-efficient orchestrator that dynamically routes queries between local and cloud models. Using Bayesian Belief Fusion and a localized experience memory, it minimizes API costs while maximizing accuracy across domain tasks.

The Open Dev

AI/ML APIAMD Developer CloudAntigravity+6

Why Incoreperated

Why Incorporated is a hybrid AI agent that classifies tasks across 8 capability domains and routes each to a local model or Fireworks AI, escalating only when local confidence is low to save tokens.

why incorporated

LLaMAGemma

AIRA (Autonomous Manufacturing Decision Platform)

AIRA is an AI-powered autonomous manufacturing decision platform that combines multi-agent intelligence, Digital Twin simulation, and AMD Developer Cloud to deliver explainable, real-time operational decisions.

Kavithai Kernels

AMD Developer CloudAMD ROCmLangChain+5

Cheap Route

CheapRoute serves LLM tasks locally on Gemma and pays for a frontier model only when needed. A confidence gate over log-probs, self-agreement, and format vetoes decides; most answers use zero remote tokens, and a wall-clock cap bounds every response.

SillyGeese

AMD Developer CloudAMD ROCmGemma+1

track1-agent:latest

ZeroRoute answers all 8 task categories with two bundled 3B models running entirely inside the container — zero Fireworks tokens by construction. Executed program-of-thought math, enumerated logic, and syntax-checked code keep accuracy high.

NGRAM

LLaMAQwen3AMD Developer Cloud

30apps loaded