It pollutes the skills list with project fragments such as 'parse 500', 'real-time sales forecasting dashboard', 'automation tool', 'aws lambda + s3', 'ci', and 'cd'.
What was measured
Noise in output
Does it add incorrect or hallucinated fields?
decisive for this rankingtransformation
Incorrect or hallucinated fields mean the resume parser is not producing reliable structured data, which is central to the job. (3 of 3 judges)
What was given, what came back
Test input: Multi-column sidebar resume — Priya Sharma · pdf · group: resume-parsing
Input — what we sent
Multi-column sidebar resume — Priya Sharma
A two-column sidebar resume for Priya Sharma, used to test whether parsers can preserve reading order and correctly extract content split across columns.
Why this input is hard
- · multi-column layout handling
- · sidebar content extraction
- · reading-order robustness
- · projects extraction
- · languages and certifications extraction
Output — unretouched
Also checked on this input — same tool, 3 other criteria
Accuracy◐ MixedOn the two-column resume, it truncates the job title, dropping the '— ML' suffix from 'Software Engineer — ML'.Multi-column handling◐ MixedIt can read a two-column/sidebar resume without crashing and pull content from both columns, but the parse is not perfectly layout-stable.Output quality◐ MixedThe two-column output is only moderate quality: core fields are present, but title truncation, noisy skills, and incomplete task extraction reduce usability.
Provenance
- Observation
- f44624ec-226f-4912-85ab-fb85254978ac
- Evidence run
- cbbef4db-964c-49fa-a57f-a2977822bdfc
- Study
- Parse resumes into structured data using an API
- Research task
- 86b9jm30n
- Tested at
- not recorded
- Source
- first-party
- Evidence state
- verified
- Proof shown
- input + output shown
- Cost / latency
- not captured
- Repeat run
- not captured
- Tester
- not captured
The last three rows are honest blanks, not placeholders — our capture has no field for them yet.
Query this
get_evidence({
tool: "hrflow",
scenario: "resume-parsing"
})MCP · mcp.aidemos.com/api/mcp
Free with attribution.
Same input, same check — 2 other tools
measured on Noise in output
Affinda✗ FailedThe skill taxonomy can hallucinate unrelated terms; on this resume it injected American Welding Society Codes even though that text is not in the source.Hireability✗ FailedThe skills output contained blank level values and mixed in non-skill labels such as Engineering and Software as competencies.
This evidence is published in
Real inputs and real outputs, no retouching · every cell queryable via API & MCP · aidemos.com