agent_decoder / TASK.md
Aziz3's picture
Initial commit: English Accent Detection Tool with Streamlit and tests
b3cdca1

A newer version of the Streamlit SDK is available: 1.49.1

Upgrade

complete the following task with sour code, explanation and referencs if available. Overview:

At REM Waste, we’re building intelligent tools to automate real hiring decisions. As part of your interview, we’d like you to complete a practical challenge that reflects the kind of work you’ll be doing here—solving real-world problems using AI tools.

Challenge Task:

Objective:

Build a working script or simple tool that can do the following:

  1. Accept a public video URL (e.g., Loom or direct MP4 link).

  2. Extract the audio from the video.

  3. Analyze the speaker’s accent to detect English language speaking candidates.

  4. Output:

  • Classification of the accent (e.g., British, American, Australian, etc.)

  • A confidence in English accent score (e.g., 0-100%)

  • A short summary or explanation (optional)

This tool will be used internally to help evaluate spoken English for hiring purposes.

What We're Looking For:

Top Priority:

  • Practicality – Can you build something that actually works?

  • Creativity – Did you come up with a smart or resourceful solution?

  • Technical Execution – Is it clean, testable, and logically structured?

You’re free to use any tools or languages you’re comfortable with (Python, JavaScript, no-code tools, open-source APIs, etc.).

Deliverables:

  • A working script, notebook, or small app (CLI, Streamlit, Flask—your choice)

  • Deploy it somewhere with simple UI so it can be tested by clicking the link

  • You can submit your work [72 hours from the receipt of this email] via this form: https://forms.gle/PTdcsAUGCKUi1BKP6.

Time Expectation:

This task is unpaid, so please don’t spend more than 4–6 hours. We’re looking for working proof-of-concept, not perfection. If you already have something similar, feel free to repurpose or expand it.

Evaluation: (Pass/Fail Screening)

Area
Must-Have for Pass Notes
Functional Script Yes Must run and return accent classification Logical Approach Yes Uses valid methods for transcription + scoring Setup Clarity
Yes Clear README to test it Accent Handling (English) Yes Only English accents are needed Bonus: Confidence Scoring Optional Points for extra polish or creativity