Pinned Loading

1. llm-token-optimization-gateway llm-token-optimization-gateway Public

   LLM gateway that cuts prompt tokens by 91.7% on a support workload through context compression and exact and semantic caching, with under 1 ms p95 overhead. Python, FastAPI, OpenAI-compatible backe…

   Python

2. nl-to-sql-agent nl-to-sql-agent Public

   Natural language to guarded, read-only SQL: AST safety checks, a self-correcting agent loop, and an execution-accuracy eval. Runs offline or with any OpenAI-compatible LLM.

   Python

3. whisperoom whisperoom Public

   Fully local voice-based mock interview system — Ollama + faster-whisper + Piper TTS. No paid APIs.

   Python