Arabic QA for Meeting Minutes
Budget: $30 – $250 USD
I need an experienced quality-assurance analyst to put our online meeting platform through its paces, with the spotlight on the Minutes of Meeting feature. Your mission is to verify that every set of minutes coming out of the system is truly publish-ready: the transcription must mirror what was said, the minutes of meeting must include every action item or follow-up mentioned in the meeting.
Because most of our calls take place in Arabic, native-level Arabic fluency is essential; you’ll be catching nuance in the conversation as well as spotting transcription slips that an automated engine might miss. I’ve already built basic checks for transcription accuracy and role-based access control, and I’d like you to validate those too, but the primary success metric is the quality of the generated minutes.
Typical workflow: join a meeting with our AI assistant, compare the raw transcript against the audio, run through the minutes produced by the platform, note all discrepancies and missing action points, then retest after fixes. Test plans, annotated screenshots, and concise defect tickets to be added in AzureDevOps and mapped to their corresponding tasks.
Deliverables:
• A test report that lists each defect or enhancement related to content accuracy, and action items/follow-ups, mapped to reproduction steps
• A short verification summary after each round of fixes confirming that issues are resolved or still open
If you have already audited meeting or transcription tools, I’d love to hear how you approached edge cases such as multiple speakers, different dialects, or restricted user roles. Turnaround is flexible but I aim to start the first test cycle this week and iterate quickly after that.
Because most of our calls take place in Arabic, native-level Arabic fluency is essential; you’ll be catching nuance in the conversation as well as spotting transcription slips that an automated engine might miss. I’ve already built basic checks for transcription accuracy and role-based access control, and I’d like you to validate those too, but the primary success metric is the quality of the generated minutes.
Typical workflow: join a meeting with our AI assistant, compare the raw transcript against the audio, run through the minutes produced by the platform, note all discrepancies and missing action points, then retest after fixes. Test plans, annotated screenshots, and concise defect tickets to be added in AzureDevOps and mapped to their corresponding tasks.
Deliverables:
• A test report that lists each defect or enhancement related to content accuracy, and action items/follow-ups, mapped to reproduction steps
• A short verification summary after each round of fixes confirming that issues are resolved or still open
If you have already audited meeting or transcription tools, I’d love to hear how you approached edge cases such as multiple speakers, different dialects, or restricted user roles. Turnaround is flexible but I aim to start the first test cycle this week and iterate quickly after that.
Related categories:
Testing / QA
Transcription
Test Automation
Arabic Translator
Documentation
Test Plan Writing