Fix artifact upload by copying result to mounted volume

- Changed workdir to /workdir (the mounted checkout directory)
- Copy result.json to /workdir before container exits
- Use uv for Python package management
- Clone from buildkite-infrastructure branch

Author: Mark Saroufim

deployment/buildkite/pipeline-eval.yml

@@ -1,10 +1,5 @@
 # Kernelbot Evaluation Pipeline for Buildkite
-# This pipeline runs real kernel evaluations by cloning the repo and running the evaluator
-#
-# To use this pipeline:
-# 1. Go to your Buildkite pipeline settings
-# 2. Paste this YAML in the Steps section
-# 3. Submit jobs via BuildkiteLauncher
+# Mirrors GitHub runner: clone repo, install deps, run evaluation
 
 steps:
   - label: ":rocket: Kernel Evaluation"
@@ -17,53 +12,56 @@ steps:
           always-pull: false
           gpus: "all"
           propagate-environment: true
+          shell: ["/bin/bash", "-e", "-c"]
           environment:
             - NVIDIA_VISIBLE_DEVICES
             - CUDA_VISIBLE_DEVICES
             - KERNELBOT_PAYLOAD
             - KERNELBOT_RUN_ID
           cpus: "${KERNELBOT_CPUS:-8}"
           memory: "${KERNELBOT_MEMORY:-64g}"
+          workdir: /workdir
 
     command: |
       set -e
 
       echo "=== Environment ==="
-      echo "NVIDIA_VISIBLE_DEVICES=$$NVIDIA_VISIBLE_DEVICES"
-      echo "KERNELBOT_RUN_ID=$$KERNELBOT_RUN_ID"
+      echo "NVIDIA_VISIBLE_DEVICES=$NVIDIA_VISIBLE_DEVICES"
+      echo "KERNELBOT_RUN_ID=$KERNELBOT_RUN_ID"
       nvidia-smi -L
 
       echo ""
-      echo "=== Installing Dependencies ==="
+      echo "=== Installing System Dependencies ==="
       apt-get update -qq
-      apt-get install -y -qq python3.11 python3.11-venv python3-pip git > /dev/null
+      apt-get install -y -qq curl ca-certificates git
 
-      # Create and activate virtual environment
-      python3.11 -m venv /tmp/venv
-      source /tmp/venv/bin/activate
+      echo ""
+      echo "=== Installing uv ==="
+      curl -LsSf https://astral.sh/uv/install.sh | sh
+      . /root/.local/bin/env
 
-      # Install PyTorch and kernelbot
-      pip install --quiet torch triton numpy scipy
+      echo ""
+      echo "=== Cloning Repository ==="
+      git clone --depth 1 --branch buildkite-infrastructure https://github.com/gpu-mode/kernelbot.git /opt/kernelbot
+      cd /opt/kernelbot
 
-      # Clone kernelbot and install
-      git clone --depth 1 https://github.com/gpu-mode/kernelbot.git /tmp/kernelbot
-      cd /tmp/kernelbot
-      pip install --quiet -e .
+      echo ""
+      echo "=== Installing Dependencies ==="
+      uv sync
 
       echo ""
       echo "=== Running Evaluation ==="
-      python src/runners/buildkite-runner.py
+      uv run python src/runners/buildkite-runner.py
 
       echo ""
+      echo "=== Copying Artifacts ==="
+      cp result.json /workdir/result.json
+      cp -r profile_data /workdir/profile_data 2>/dev/null || true
+
       echo "=== Done ==="
 
     artifact_paths:
       - "result.json"
-      - "profile_data/**/*"
+      - "profile_data/*"
 
     timeout_in_minutes: 30
-
-    retry:
-      automatic:
-        - exit_status: -1
-          limit: 2