What is included?
Reports cover active U.S. software engineering jobs in the CronJobs production inventory at one observation time. A job must have a known role category, a working source or employer application URL, an in-scope software-engineering classification (or the explicit legacy compatibility marker), and a source confirmation within the prior seven days.
The production ingestion pipeline owns the conservative U.S.-location and software-engineering filters. Reporting can narrow that eligible set but does not reclassify rejected jobs or broaden geographic scope.
Which measures are published?
| Measure | Definition |
|---|---|
| Active jobs | Distinct eligible job records at the observation time. |
| Companies | Distinct normalized company labels represented by eligible jobs. |
| Role mix | Eligible jobs across AI/ML, backend, full-stack, frontend, DevOps/SRE, mobile, and security. |
| Work mode | Normalized remote, hybrid, onsite, or unknown classification supplied by the pipeline. |
| Seniority | Normalized junior, mid, senior, staff, principal, lead, or unknown classification. |
| Listing age | Time since the source posting date, falling back to CronJobs first-seen time when the source omits its date. |
| Salary disclosure | Share with at least one source-derived salary bound. It is not a salary-level estimate. |
| Source mix | Each job assigned to its earliest stored source so duplicate ingestion paths do not inflate totals. |
| Technology demand | Distinct eligible jobs containing a normalized technology tag. |
Freshness and duplicates
CronJobs refreshes sources throughout the day and marks a job inactive after seven days without confirmation. Canonical job keys deduplicate the same opening across repeated crawls and multiple sources. A source correction, removal, or republication can therefore change a later snapshot.
Privacy and publication thresholds
Public report files contain aggregate job-market data only. They exclude Discord identities, preferences, messages, resumes, interactions, individual salary submissions, and raw language-model fit fields. Technology and source rows require at least ten eligible jobs; salary-disclosure rows by role also require at least ten jobs.
Known limitations
- The dataset describes the CronJobs source registry, not every software engineering opening in the United States.
- Classification is conservative and can leave ambiguous role, seniority, or work-mode values unknown.
- A listed salary can come from source text normalization; reports do not treat it as a verified compensation outcome.
- Counts are point-in-time observations and should not be compared without distinct observation timestamps and the same methodology version.
- Changes in source coverage can affect trends independently of the labor market.
Downloads and citation
Each published report receives a dated directory containing a machine-readable JSON snapshot, stable CSV tables, and a README repeating the observation time and limitations. The undated manifest points to the newest completed snapshot; prior dated files remain immutable.
Read the September 2026 report, download the latest JSON snapshot, or inspect the dataset manifest.
Cite a snapshot as: CronJobs U.S. Software Engineering Jobs data snapshot, followed by its exact observation timestamp, dated snapshot URL, and this methodology URL. Do not cite the undated “latest” pointer without recording the observation timestamp it returned.
Reproduction and corrections
The checked-in report generator uses aggregate, read-only D1 queries and publishes its schema and definitions with every snapshot. If a material query or definition changes, CronJobs advances the methodology version instead of silently overwriting the old contract. Corrections to a published snapshot receive a documented replacement rather than an in-place unexplained edit.