diff --git a/.github/workflows/ci-validate-challenge-config.yml b/.github/workflows/ci-validate-challenge-config.yml
new file mode 100644
index 000000000..c3c04728c
--- /dev/null
+++ b/.github/workflows/ci-validate-challenge-config.yml
@@ -0,0 +1,69 @@
+name: CI - Validate challenge_config
+
+on:
+  pull_request:
+    branches: [ master ]
+    paths:
+      - "challenge_config.yaml"
+      - "challenge_config.yml"
+      - "evaluation_script/**"
+      - "challenge_data/**"
+      - "github/**"
+  push:
+    branches: [ master ]
+    paths:
+      - "challenge_config.yaml"
+      - "challenge_config.yml"
+      - "evaluation_script/**"
+      - "challenge_data/**"
+      - "github/**"
+
+permissions:
+  contents: read
+
+jobs:
+  validate:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout PR code / pushed commit
+        uses: actions/checkout@v4
+
+      - name: Set up Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: "3.11"
+
+      - name: Install dependencies
+        run: |
+          python -m pip install --upgrade pip
+          if [ -f github/requirements.txt ]; then
+            pip install -r github/requirements.txt
+          fi
+          pip install pyyaml
+
+      - name: Pick challenge config path
+        id: pick
+        run: |
+          set -e
+          if [ -f challenge_config.yaml ]; then
+            echo "cfg=challenge_config.yaml" >> $GITHUB_OUTPUT
+          elif [ -f challenge_config.yml ]; then
+            echo "cfg=challenge_config.yml" >> $GITHUB_OUTPUT
+          else
+            echo "::error::challenge_config.(yaml|yml) not found"
+            exit 1
+          fi
+
+      - name: Validate challenge config (offline checks)
+        run: |
+          python github/ci_validate_challenge_config.py "${{ steps.pick.outputs.cfg }}"
+
+      # Optional: only on push to master, run EvalAI API validation if secrets exist
+      - name: Validate with EvalAI API (optional, master only)
+        if: github.event_name == 'push'
+        env:
+          EVALAI_AUTH_TOKEN: ${{ secrets.EVALAI_AUTH_TOKEN }}
+          EVALAI_HOST_TEAM_PK: ${{ secrets.EVALAI_HOST_TEAM_PK }}
+          EVALAI_HOST_URL: ${{ secrets.EVALAI_HOST_URL }}
+        run: |
+          python github/ci_validate_challenge_config.py "${{ steps.pick.outputs.cfg }}"
diff --git a/github/ci_validate_challenge_config.py b/github/ci_validate_challenge_config.py
new file mode 100644
index 000000000..64da140d0
--- /dev/null
+++ b/github/ci_validate_challenge_config.py
@@ -0,0 +1,143 @@
+import http
+import os
+import sys
+import zipfile
+from pathlib import Path
+
+import requests
+
+try:
+    import yaml  
+except ImportError:
+    yaml = None
+
+
+REPO_ROOT = Path(__file__).resolve().parents[1]
+
+
+CONFIG_CANDIDATES = [
+    REPO_ROOT / "challenge_config.yaml",
+    REPO_ROOT / "challenge_config.yml",
+]
+
+
+IGNORE_DIRS = {".git", ".github", ".venv", "__pycache__", "node_modules"}
+IGNORE_FILES = {"challenge_config.zip", "evaluation_script.zip"}
+
+
+def find_config_file() -> Path:
+    for p in CONFIG_CANDIDATES:
+        if p.exists():
+            return p
+    raise FileNotFoundError("challenge_config.(yaml|yml) not found at repo root")
+
+
+def yaml_sanity_check(config_path: Path) -> None:
+    if yaml is None:
+        raise RuntimeError("PyYAML not installed. CI should install pyyaml before running this script.")
+    with config_path.open("r", encoding="utf-8") as f:
+        yaml.safe_load(f)
+
+
+def build_evaluation_script_zip(repo_root: Path) -> None:
+    eval_dir = repo_root / "evaluation_script"
+    if not eval_dir.exists():
+        # Some CI runs may not need it; keep it non-fatal.
+        print("INFO: evaluation_script/ not found; continuing.")
+        return
+
+    out_zip = repo_root / "evaluation_script.zip"
+    with zipfile.ZipFile(out_zip, "w", zipfile.ZIP_DEFLATED) as zf:
+        for file_path in eval_dir.rglob("*"):
+            if file_path.is_file():
+                zf.write(file_path, file_path.relative_to(eval_dir))
+    print("INFO: Built evaluation_script.zip")
+
+
+def build_challenge_config_zip(repo_root: Path) -> Path:
+    out_zip = repo_root / "challenge_config.zip"
+    if out_zip.exists():
+        out_zip.unlink()
+
+    with zipfile.ZipFile(out_zip, "w", zipfile.ZIP_DEFLATED) as zf:
+        for path in repo_root.rglob("*"):
+            rel = path.relative_to(repo_root)
+
+            if any(part in IGNORE_DIRS for part in rel.parts):
+                continue
+            if path.name in IGNORE_FILES:
+                continue
+            if path.is_file():
+                zf.write(path, rel)
+
+    return out_zip
+
+
+def evalai_validate(zip_path: Path) -> bool:
+    """
+    Uses ONLY the validate API
+    If secrets are missing, we skip API validation but keep YAML sanity checks.
+    """
+    token = (os.getenv("EVALAI_AUTH_TOKEN") or "").strip()
+    team_pk = (os.getenv("EVALAI_HOST_TEAM_PK") or "").strip()
+    host_url = (os.getenv("EVALAI_HOST_URL") or "https://eval.ai").strip().rstrip("/")
+
+    if not token or not team_pk:
+        print("INFO: EVALAI_AUTH_TOKEN / EVALAI_HOST_TEAM_PK not set -> skipping EvalAI API validation.")
+        return True
+
+    url = f"{host_url}/api/challenges/challenge/challenge_host_team/{team_pk}/validate_challenge_config/"
+    headers = {"Authorization": f"Bearer {token}"}
+
+    with zip_path.open("rb") as f:
+        files = {"zip_configuration": f}
+        data = {"GITHUB_REPOSITORY": os.getenv("GITHUB_REPOSITORY", "")}
+        resp = requests.post(url, headers=headers, files=files, data=data, timeout=120)
+
+    if resp.status_code in (http.HTTPStatus.OK, http.HTTPStatus.CREATED):
+        print("✅ EvalAI API validation passed.")
+        return True
+
+    # Best-effort error printing
+    try:
+        payload = resp.json()
+    except Exception:
+        print(f"❌ EvalAI validation failed: HTTP {resp.status_code}\n{resp.text}")
+        return False
+
+    detail = payload.get("detail")
+    error = payload.get("error")
+    if detail:
+        print(f"❌ EvalAI validation failed: {detail}")
+    if error:
+        print("❌ Following errors occurred while validating the challenge config:\n{}".format(error))
+    if not detail and not error:
+        print(f"❌ EvalAI validation failed: HTTP {resp.status_code}\n{payload}")
+
+    return False
+
+
+def main() -> int:
+    config_path = find_config_file()
+
+    print(f"INFO: Found config: {config_path.name}")
+    yaml_sanity_check(config_path)
+    print("✅ YAML syntax check passed.")
+
+    
+    build_evaluation_script_zip(REPO_ROOT)
+    zip_path = build_challenge_config_zip(REPO_ROOT)
+    print(f"INFO: Built challenge zip: {zip_path.name}")
+
+    ok = evalai_validate(zip_path)
+
+    
+    for p in [REPO_ROOT / "evaluation_script.zip", REPO_ROOT / "challenge_config.zip"]:
+        if p.exists():
+            p.unlink()
+
+    return 0 if ok else 1
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())
diff --git a/github/requirements.txt b/github/requirements.txt
index ed667b853..7416b4b3e 100644
--- a/github/requirements.txt
+++ b/github/requirements.txt
@@ -1,2 +1,3 @@
+PyYAML==6.0.3
 PyGithub===1.53
 requests==2.32.4