Ranked Award Thesis
| Rank | Award route | Why this is strongest | Evidence |
|---|---|---|---|
| 1 | Blog Post Award | The story is the product loop: PrizePilot uses Qwen to choose a realistic cash-prize route, build public proof, and stop at human approval gates instead of inventing private/account evidence. | Blog story, share packet, before/after evidence, Qwen contribution map |
| 2 | Top 10 Honorable Mention Projects | The submission is public, reproducible, and clear about completed evidence versus pending sponsor proof. | judge pack, evidence map, cloud readiness, judge manifest |
| 3 | Track 4 Autopilot Agent fit | The core product is an autopilot agent for route triage, artifact planning, and human gate management. | src/prizepilot/agent.py, src/prizepilot/qwen_client.py, static plan JSON, tests |
Judge-Visible Strengths
- Public submission identity: Devpost, GitHub repo, Vimeo demo, Pages hub, deck, static demo, and public blog are reviewable without private account access.
- Qwen contribution: live Qwen/DashScope smoke proof exists, and before/after evidence shows how Qwen sharpens route language, story clarity, and risk wording.
- Reproducibility: deterministic planning still works without a model key, so Qwen refinement improves the narrative without hiding the baseline.
- Integrity boundary: Alibaba endpoint proof remains pending until a real public URL returns HTTP 200 at
/and/api/plan.
Do Not Infer
Boundary: this scorecard is an award-thesis summary, not a statistical prediction of winning and not a prize, payout, tax, KYC, billing, bank, or cloud-credit claim.
- Do not infer PrizePilot has won, been shortlisted, or received a payout.
- Do not infer a live Alibaba Cloud public endpoint exists until public HTTP proof is captured.
- Do not publish API keys, request headers, account IDs, cookies, private console screenshots, private email, payment data, or localStorage.
Verification
Local verification remains the existing no-secret test path: $env:PYTHONPATH='src'; python -m unittest discover -s tests -v, python -m compileall -q src tests tools, python -m prizepilot qwen-status, python -m prizepilot cloud-readiness, and python -m json.tool docs\judge-manifest.json.