Skip to content

Job Analyze API

POST /api/v1/job/analyze

Use this endpoint when an application needs ranked occupation candidates, supporting evidence, explanations, and decision-support fields from job text.

/api/v1/job/analyze is the highest-context job endpoint. Use it when your product needs to compare candidate occupations, expose skill evidence, show why a match was selected, inspect related occupation context, or route results into review and automation using confidence, trust, uncertainty, ambiguity, and weak-signal fields.

Complete executable examples belong in the AvelinLabs API examples repository.

Base URL: https://api.avelinlabs.com

Requires a runtime API key:

Authorization: Bearer <runtime-api-key>

Type: string

Required: Yes

Meaning: Job title or short role label to analyze.

Type: string

Required: No

Meaning: Job description or other skill-bearing text. More complete text can improve the quality of evidence.

Type: boolean

Required: No

Meaning: Optional diagnostic flag. Public integrations should not depend on debug-only fields.

{
"title": "Data Analyst",
"description": "Build dashboards, analyze business metrics, create reports, and communicate insights to stakeholders.",
"debug": false
}

See the public example payload: payloads/job-analyze.json.

Illustrative beta response snippet:

{
"job_title": "Data Analyst",
"confidence": 0.88,
"trust_score": 0.84,
"confidence_level": "HIGH",
"job_signals": ["dashboards", "reporting"],
"uncertainty": {
"total": 0.185
},
"decision": {
"decision": "AUTO_ACCEPT",
"reason": "Confidence and uncertainty meet the configured decision threshold."
},
"results": [
{
"onet_code": "15-2051.01",
"title": "Business Intelligence Analysts",
"confidence": 0.88,
"matching_skills": ["Microsoft Power BI"],
"explanation": {
"why_match": "The role aligns with dashboarding, reporting, and analytical work."
}
}
]
}

See the public example response: responses/job-analyze.example.json.

Meaning: Echoes or normalizes the job title used for analysis.

Meaning: Version label for the response-producing model or scoring configuration. Treat as metadata.

Meaning: Top-level confidence signal for the overall result. Use as decision support, not absolute truth.

Meaning: Raw or pre-adjusted confidence signal where present. Most clients should prefer confidence and confidence_level.

Meaning: Overall trust-oriented score derived from result strength and quality support.

Meaning: Categorical confidence label such as HIGH, intended for display and routing logic.

Meaning: Optional beta-era categorical confidence field. Clients should handle it if present and ignore it if absent.

Meaning: Describes the semantics of the top-level confidence value.

Meaning: Describes how per-result confidence values are represented.

Meaning: Object containing uncertainty signals. See output interpretation.

Meaning: Optional decision block that can help route results to automation or review.

Meaning: Broad inferred task or role category when available.

Meaning: Skill, tool, task, or domain signals detected from the input.

Meaning: Structured intent or archetype information when available. Treat as supporting context.

Meaning: Signal indicating how reliable the title appears as evidence.

Meaning: O*NET family codes associated with the result set.

Meaning: Signal for how strongly the input appears to fit the supported domain.

Meaning: Categorical domain support label when present.

Meaning: Human-readable summary of the domain support signal.

Meaning: Indicates whether the domain evidence is ambiguous.

Meaning: Indicates whether the overall result should be treated as ambiguous.

Meaning: Indicates that available evidence may be sparse or weak.

Meaning: Short reason for weak-signal detection when present.

Meaning: Optional metadata indicating whether model-assisted reasoning was used. Do not build core workflow logic around these fields.

Meaning: Optional explanation text when available. Treat as supporting narrative, not a contract guarantee.

Meaning: Ranked array of occupation matches.

Meaning: O*NET occupation code for the candidate result.

Meaning: Broader O*NET family code.

Meaning: Occupation title.

Meaning: Per-result confidence within the returned ranked set. Interpret relative to other returned candidates.

Meaning: Signal for whether the result is plausible for the input.

Meaning: Aggregate quality signal for the candidate.

Meaning: Additive object of numeric quality components. Clients should ignore unknown keys.

Meaning: Categorical support for the result within the supported domain.

Meaning: Categorical alignment signal.

Meaning: List of potential quality concerns, if any.

Meaning: Suggestions for improving input quality, if available.

Meaning: O*NET 30.3-grounded skill records, typically including skill name, importance, and level.

Meaning: Knowledge-area records associated with the occupation.

Meaning: Ability records associated with the occupation.

Meaning: Example task statements associated with the occupation.

Meaning: Technology records associated with the occupation, optionally including demand or hot-technology flags.

Meaning: Skills from the input or context that support the match. These may not exactly match O*NET skill labels.

Meaning: Indicates how matching skills were derived, for example from extracted input signals.

Meaning: Related occupation suggestions with O*NET codes, titles, tiers, and scores where available.

Meaning: Structured explanation object.

Meaning: Optional confidence explanation text.

Meaning: Human-readable reason the occupation is a match.

Meaning: List of important evidence points.

Meaning: Object with score components such as semantic match, skill match, domain alignment, and signal overlap where available.

Use ranked results and confidence fields as decision-support signals. Ambiguous, weak-signal, or low-confidence outputs should generally be reviewed before automation.

Use /job/classify instead when your integration only needs one top O*NET occupation and does not need the ranked candidate list or expanded evidence.

For cross-endpoint guidance, read output interpretation and standard error responses.