Harkanwal Bedi

Projects & Open Source

Merged upstream contributions, plus reference architectures and tooling for Vertex AI and Gemini.

Merged pull requests

  • Merged C++

    ggml-org/llama.cpp #30779 

    llama-server now visits parent slots before their child slots, so requests that share a prompt prefix no longer need a third forward pass. Released in b12188.

  • Merged Python

    AI-Hypercomputer/maxdiffusion #476 

    Skips the CUDA-only nvidia-cuda-cccl dependency on macOS, where no wheels exist, so local installs no longer fail on Apple Silicon.

Repositories

All on GitHub

  • Terraform · Vertex AI

    vertex-llm-benchmark 

    Terraform and pipeline harness to benchmark LLM serving latency, throughput, and capacity sizing on Vertex AI.

  • MLOps · Vertex AI

    continuous-training-vertex-ai 

    End-to-end Continuous Training (CT) MLOps pipeline on Google Cloud Vertex AI.

  • Python · Gemini

    pdf-wizard 

    Toolkit and evaluation suite for automated AcroForm PDF filling and structured document workflows with Gemini.

  • Python · Gemini

    gemini-model-availability-checker 

    Verify Gemini model and regional availability across Google AI Studio, Vertex AI, and Gemini Enterprise.

  • Jupyter · Vertex AI

    vertexai-llm-notebooks 

    Applied Vertex AI notebooks for RAG, embedding dimension tuning, text-to-SQL, and Enterprise Search with Gemini.