MUDRAIS: Semantic Matchmaking Engine
MUDRAIS (Multi-User Dynamic Roleplay AI System) solves a fundamental problem in online communities: finding truly compatible collaborators. Traditional bots rely on keyword filters that miss nuance — narrative tone, writing style, thematic compatibility. MUDRAIS replaces that with a semantic matchmaking engine. When a user registers, their free-text input enters a multi-agent AI pipeline running on AMD MI300X via AMD Developer Cloud. A Gatekeeper agent (OSS 20B) extracts structured fields from natural language. A Context Optimizer agent (OSS 120B) generates a semantically dense representation tuned for embedding. The nvidia/llama-nemotron-embed-vl-1b-v2 model then converts it into a 2048-dimensional float vector stored in Qdrant. The matchmaking engine runs cosine similarity search inside Qdrant, applying hard payload filters for "red lines" (absolute blockers) and strict guild-level tenant isolation — users from Server A mathematically cannot appear in Server B results. A secondary scoring layer applies soft penalties (yellow lines, timezone mismatch) on top of raw cosine distance to produce a final compatibility score. The system is Discord-agnostic by design. Discord is just the transport layer (slash commands, webhooks). The entire semantic pipeline — profile ingestion, multi-agent optimization, vector indexing, and partner search — lives in a clean Domain-Driven Design architecture on Laravel. The same engine can power Slack bots, web forms, or REST APIs without touching domain code. Each community is an independent tenant with isolated Qdrant payloads, its own archetype configuration, and its own LLM prompts — true B2B multi-tenancy at scale.
Category tags:.png&w=96&q=75){kind=small}
"🌟 Why Is It Important? - Better matching for Discord servers - Revolution for roleplay communities - People meet truly compatible people - Can be sold as B2B (to many servers) This addresses a real problem in online communities - finding truly compatible collaborators. Traditional keyword-based bots miss nuance like narrative tone and writing style compatibility. MUDRAIS replaces this with a sophisticated semantic matchmaking engine using multi-agent AI and vector embeddings. The domain-driven design and true multi-tenancy architecture make it enterprise-ready. Application of Technology: 🚀🚀🚀🚀🚀 5 - Multi-agent pipeline with Gatekeeper (20B) and Context Optimizer (120B). Vector embedding using nvidia/llama-nemotron-embed-vl-1b-v2 → 2048-dim vectors in Qdrant. Cosine similarity search with hard payload filters and tenant isolation. AMD MI300X via AMD Developer Cloud. Domain-Driven Design on Laravel. Sophisticated architecture. Presentation: 🚀🚀🚀🚀 4 - Clear problem statement (keyword filters miss nuance). Good explanation of the pipeline and architecture. Links to GitHub and Discord provided. Well-structured description. Business Value: 🚀🚀🚀🚀🚀 5 - Massive online community market. Roleplay communities, Discord servers all need better matching. True B2B multi-tenancy allows scaling to many communities. Could be adapted for other platforms beyond Discord. Originality: 🚀🚀🚀🚀🚀 5 - Very original. Semantic matchmaking with vector embeddings is innovative. The tenant isolation with mathematical guarantees is clever. Multi-agent optimization of user profiles is unique. First-of-its-kind for roleplay community matching. "
Sanem Avcil
For Innovators & Creators