Scoring and Limits
How Round 2 answers are scored, and the submission limits
Scoring and Limits
| # | Challenge | Points | Scored by |
|---|---|---|---|
| 1 | Equipment Catalog | 100 | F1 over each facility's pairs with counts, plus the most common model |
| 2 | Shared Models | 100 | F1 over the shared pairs, plus count accuracy |
| 3 | Standardization Dictionary | 200 | Grouping, Partner Hospital spellings and counts |
| 4 | Device Types and Useful Life | 200 | Per-asset device type and useful life |
| 5 | Capital Projection | 150 + 50 | Per-year replacement schedule; bonus for per-asset cost estimates |
| 6 | GPO Recommendation | 150 | The Equiply team, reading your pick and reasoning |
Total: 900 points, plus 50 bonus. Your best score on each challenge counts toward the leaderboard.
The questions build on each other. Q1 and Q2 work on the names exactly as exported. Q3 standardizes them. Q4's device types and useful lives drive the Q5 schedule, and the Q5 schedule drives the Q6 recommendation.
How Answers Are Matched
- Raw names (Q1, Q2 and the raw spellings in Q3): compared ignoring case and runs of whitespace. Nothing else is normalized.
- Standard names (Q3 keys): never compared with ours. Only the grouping counts: spellings of the same device should share one key, different devices shouldn't.
- Device types (Q4, Q6): must be a
typefrom the Device Types API, compared ignoring case, spacing and plurals. - Assets (Q4, Q5): identified by
CEID; Q4 also usesSerial Numberwhen a CEID isn't enough.
Write-ups
Q3 to Q6 ask for something in writing, and each one is required:
- Q3 and Q4: a process write-up in
process. - Q5: a report in
report. - Q6: your reasoning in
reasoning.
Write-ups are read by the Equiply team; they don't change your score on Q3 to Q5. Q6 is scored by the team as a whole, reasoning included. Write-ups are markdown, up to 12,000 characters, and each solution page lists what we look for. Specific beats general: name the problems you hit and show how you dealt with them.
Write them yourself, in your own words. You can use AI tools for the work, but the write-up should be yours: we want to hear how you think.
Feedback
Responses report your score and the accuracy of each scored part. They never say which records or pairs were wrong. Q6 responses only confirm receipt, with "status": "pending_review".
Limits
- 5 submissions per challenge. The limit doesn't reset.
- After your fifth submission the challenge is locked and your best score stands. Each response includes
attempts_remaining. - Invalid requests (any
400) don't use a submission. Every valid one does, Q6 included. - Data and device-type endpoints: 60 requests per minute per API key.