Skip to content
Labeling Jobs

AI Safety Experts: English & Odia

Pay
$16 – $22 / Hour
Open to
Worldwide
Apply

We earn a commission if you sign up through the links on this page. It costs you nothing and does not affect which jobs we list. How this works.

Skills
  • red teaming
  • ai response evaluation
  • content review
  • low-resource languages
  • odia
  • data annotation

What you'll do

Odia (spoken mainly in Odisha) is among the lowest-resource languages Mercor is testing in this AI safety family. Models see very little Odia text in training, so their Odia output is often thin, inaccurate or borrowed from Bengali or Hindi, and their safety behaviour in Odia is largely untested. That scarcity is likely why this listing has hired more people this month (88) than any other Indian-language version we saw.

The ad describes red-team tasks:

  • Try to break chat models and agents using jailbreaks, prompt injections, misuse cases, bias exploitation and multi-turn manipulation
  • Annotate failures, classify vulnerabilities and flag systemic risks
  • Apply the project's taxonomies, benchmarks and playbooks
  • Document findings in reproducible reports, datasets and attack cases

The candidate profile for Odia centres on judgment rather than security experience. Mercor wants people who can tell whether an AI response is accurate, complete and appropriate, notice subtle errors, work consistently to guidelines and explain their reasoning. For a low-resource language, that often means first establishing whether the model understood the prompt at all before judging whether its answer is safe.

Everything is text. Sensitive topics (bias, misinformation, harmful behaviour) are included; topics are communicated first, and higher-sensitivity projects are optional with guidelines and wellness resources.

Who fits

  • Native fluency in Odia and English
  • Sound judgment on AI answer quality, with clear reasons
  • Attention to detail and consistent application of guidelines
  • Good written communication
  • Flexibility across tasks and projects

Nice to have: adversarial ML, cybersecurity, harassment and disinformation analysis, psychology, acting or writing. No degree or minimum years are stated.

What it pays

$16–22 per hour, Mercor's published figure and the same band as the other Indian-language variants and Urdu. The ad states weekly payment via Stripe or Wise; H-1B and STEM OPT candidates are not supported.

Scarcity does not lift the rate here: Odia speakers are harder to find than Hindi-belt speakers, but the band is identical.

Worth knowing

Good:

  • 88 hires this month, the most among Mercor's Indian-language variants when we checked
  • Rare-language skill that few annotation platforms actively recruit for
  • Evaluation-focused profile; security experience is a bonus
  • Weekly US-dollar payouts, no residence requirement published

Less good:

  • Lowest pay band in the family despite the language's scarcity
  • Adversarial tasks and distressing content are part of the work
  • No weekly hours stated; projects can end early
  • Independent contractor, clients unnamed

About this listing

A Mercor remote hourly contract, read from the Mercor site on 24 September 2026. Pay, requirements, payment terms and the visa exclusion come from the ad. Mercor states the work involves no confidential information from any employer, client or institution. See Mercor.

More roles at Mercor

See all 304

Guides about Mercor

See all 52

Browse similar roles

Not the right fit?

See every open role, or get new ones on Telegram or Discord as they are added.