Smart LLM Router
Cost-optimized intelligent LLM routing system that dynamically routes prompts based on complexity.
- ▸Prompt classification engine using heuristic-based routing logic (token count, keywords, prompt length, and query complexity).
- ▸Integrated lightweight Hugging Face models for simple queries and Google Gemini for advanced reasoning tasks.
- ▸Modular backend architecture with separate routing, classification, and model abstraction layers.
- ▸API-based inference pipelines with secure environment variable management and read-only token authentication.
- ▸Real-time response analytics: token estimation, latency tracking, model tracing, and estimated API cost monitoring.
- ▸Interactive Streamlit UI to visualize routing decisions, model selection, and generated responses.
