Pinned Loading

1. Inference-Benchmarks Inference-Benchmarks Public

   InferBench — FastAPI + ONNX Runtime inference server with a six-scenario benchmark harness: dynamic batching, caching, INT8 quantization, backpressure. INT8: 240 → 670 req/s on CPU.

   Python 2

2. greenleaf-dashboard greenleaf-dashboard Public

   2nd place, RBC × BCCAI × SFU Beedie Agribusiness Analytics Hackathon — calibrated plant-stress early-warning model (0.78 ROC-AUC) + agentic RAG analyst with cited sources

   Python 1 1

3. genai-stack-coder genai-stack-coder Public

   Fine-tuning Qwen2.5-Coder-7B (QLoRA, one 12GB GPU) to write current OpenAI/Anthropic SDK code: 12% → 72% on a 50-prompt benchmark. Data harvest → train → GGUF/Ollama → eval → agent demo.

   Python

4. mailo mailo Public

   Multi-agent email-campaign system on Amazon Bedrock AgentCore — A2A + MCP, zero-trust identity, Cedar policies, LLM-as-judge evals

   Python

5. strata-warehouse strata-warehouse Public

   End-to-end SaaS subscription analytics warehouse: Postgres → dlt → MinIO Parquet → dbt-duckdb → Dagster → Streamlit, gated by 179 dbt tests

   Python

6. analyst-agent analyst-agent Public

   An AI analyst for messy data: diagnoses, cleans, and answers questions over your tables, re-checking every fix and answer before it counts.

   Python