Agent objective
- Interact with the supplied stateful environment.
- Produce verifier-checkable actions or artifacts.
- Maximise scalar reward under the package contract.
A demanding data-engineering repair task for incremental event-time reconciliation across thousands of interleaved streams. Timestamps can expand into timezone and daylight-saving candidates, while exact elapsed constraints, cross-chunk state, ambiguity witnesses, deterministic output, and bounded memory must remain correct at up to 250,000 events without enumerating complete histories.
A demanding data-engineering repair task for incremental event-time reconciliation across thousands of interleaved streams. Timestamps can expand into timezone and daylight-saving candidates, while exact elapsed constraints, cross-chunk state, ambiguity witnesses, deterministic output, and bounded memory must remain correct at up to 250,000 events without enumerating complete histories.
Enough detail to understand the intellectual terrain; generated instances, hidden mechanisms, and solution paths remain inside the private package.
| Environment | Mathematical or technical frontier | Adaptive research problem |
|---|---|---|
| Timezone expansion | IANA zones and daylight-saving folds | Normalize naive and aware timestamps into deterministic candidate sets while respecting ambiguity and fold semantics. |
| Exact transition graph | Integer-nanosecond elapsed constraints | Propagate candidate histories across inclusive minimum/maximum bounds without floating-point time errors. |
| Streaming state | Interleaved append-only chunks | Preserve per-stream graph state across chunks without retaining full input DataFrames. |
| Ambiguity and commitment | Multiplicity-preserving dynamic programming | Keep distinct histories after convergence and commit a prefix only when every survivor agrees on candidate and elapsed time. |
| Scalable deterministic output | Witnesses, schema, canonical state | Classify open, unique, ambiguous, or impossible streams with exact pandas contracts at up to 250,000 events. |
The included note records a 55-minute GPT-5.6 solution within a 75-minute horizon. It passed all four public tests, 8,500 independent randomized checks, and two 200,000-event local scale cases, but was not submitted to the official verifier because Podman and episode credentials were unavailable. Seller QA separately shows the reference at 101/101; eight of eleven flawed mutants passed every public test but were caught privately.
We publish aggregate behavior and task structure, while withholding generated instances, hidden labels, exact successful probes, private checks, and solution trajectories.
Shown with its provenance and limitations; it is not a performance guarantee.
The supplied notes record solution time only; no normalized public reward was supplied.
As identified by the supplied artifact.
1 reported runs.
notes.md
Machine-readable provenance and the exact displayed metric are available in results.json.
The paid ZIP will live in a private R2 bucket. Vercel authorizes the buyer and issues a 2–5 minute object URL; R2 serves the bytes directly.
Authenticated buyer + entitlement check
+ private R2 object + 2–5 minute signed URL
= direct, auditable download
Package SHA-256
ff109e5fa79530521e53b07a8f9e265ad1f23944f51a90722ed8542c2b020ebeOne purchase licenses this identified item to one legal organisation for worldwide, perpetual commercial model training, evaluation, research and development. Redistribution and resale of the package are not permitted.