Jwumasay joo-ma
← All roles
Rlhf · Onsite

RLHF Preference Annotator (Onsite, Manila)

Corpshore AI's onsite RLHF Preference Annotator role in Manila, Philippines pays annotators to compare AI model outputs against a rubric and produce ranked preference data for LLM alignment, on an hourly contract with no AI background required.

Manila, PhilippinesOnsite · ContractEnglish, Filipino$5–$10/hr (indicative)UTC+8
Role overview

Manila is one of Corpshore AI's core English-language delivery hubs. As an onsite RLHF Preference Annotator, you'll compare pairs (or sets) of AI model responses against a structured rubric and produce the ranked preference data that foundation-model teams use to align large language models — the same category of work behind Corpshore AI's 4.2M-preference-pair alignment programs.

Key responsibilities

  • Compare model outputs against a written rubric and record structured preference judgments in Jwuma
  • Write concise rationale notes justifying each ranking decision
  • Identify unsafe, biased, or low-quality model outputs and flag them per project guidelines
  • Maintain consistency with gold-standard calibration sets
  • Meet daily/weekly batch targets without sacrificing judgment quality

What we look for

  • Excellent English reading comprehension and written communication
  • Strong, consistent judgment when comparing nuanced or subjective content
  • Comfortable with sustained, screen-based cognitive work
  • Available onsite at the Manila hub on a standard shift schedule
  • No AI/ML background required — rubric and platform training provided

Nice to have

  • Background in writing, editing, journalism, teaching, or research
  • Prior BPO, QA, or content-review experience
  • Exposure to chatbots or generative AI tools as a user

Compensation and benefits

  • Hourly contract pay via Jwuma, weekly payout
  • Paid rubric and platform onboarding
  • Growth path to Senior RLHF Annotator or QA Lead
  • Onsite facilities at the Manila delivery hub

Common questions

What is an RLHF preference annotator and what do they do?

They compare pairs of AI model outputs against a rubric and rank which response is better, producing the labeled preference data used to align large language models through reinforcement learning from human feedback.

Do I need a technical background to work in RLHF at Corpshore AI?

No. Strong reading comprehension, written English, and sound judgment matter most; the rubric and Jwuma platform are trained on the job.

Related

Similar roles

Apply once, get matched to what fits

One application covers every role you qualify for. A person reads it.

Apply now