Pinned Loading

1. meridian meridian Public

   Real-time AI backend built on Kafka, FastAPI, and pgvector. Includes a ReAct agent loop, multi-layer memory, multi-model routing, and full observability stack.

   Python

2. sentinel sentinel Public

   Four FastAPI microservices that sit between your app and any LLM: PII detection, prompt injection blocking, async eval, and CUSUM drift detection.

   Python

3. forge forge Public

   Multi-agent incident response system. Describe a production alert and it investigates logs, git history, and known failure patterns to produce a structured postmortem.

   Python

4. realtime-feature-platform realtime-feature-platform Public

   Feature platform that makes training-serving skew structurally impossible: shared compute path, point-in-time joins, version guards, and PSI skew detection

   Python

5. streaming-lakehouse-optimizer streaming-lakehouse-optimizer Public

   Streaming lakehouse (Postgres → Debezium → Kafka → Flink → Iceberg) with a learned, guarded layout optimizer

   Python

6. local-llm-eval local-llm-eval Public

   Evaluation harness and fallback gateway for choosing locally hosted LLMs (Qwen, Llama, Mistral on Ollama) for network-operations RAG and tool calling

   Python