Most pages ranking for AI jobs in the Philippines recycle the same global role list with no local evidence. This report does something simpler. On 6 August 2026 we pulled live posting counts for six AI-adjacent roles from LinkedIn's public Philippine job search, then cross-checked pay and industry figures against published sources. Every number below links to the page we took it from, and every count carries its pull date.

The pattern is specific. Hiring is strongest in roles that sit beside AI systems, checking them, operating them, and feeding them data. The employers posting these jobs are not distant research labs. They are BPO and shared-services firms Filipinos already know.

The posting counts, role by role

All counts come from LinkedIn's Philippine job search, retrieved 6 August 2026. Each figure is the total that the public search page reported for that keyword.

  • Prompt engineer: 631 postings, including LLM Engineer roles at MicroSourcing and KDCI Outsourcing and an AI Prompt Engineer role at QBE Group Shared Services Centre.
  • Artificial intelligence (general keyword): 508 postings.
  • AI operations specialist: 369 postings, with listings from Globe Telecom, Concentrix, and Shopee.
  • AI trainer: 225 postings, including an SLM Trainer at DysrupIT and AI Content Evaluator roles at Innodata.
  • Conversation designer: 186 postings, a looser match that pulls in chatbot analyst and instructional design roles at Shopee, Accenture, and TTEC.
  • Data annotation: 134 postings, from Sutherland, TaskUs, Capital One, and Innodata.
  • AI quality assurance: 1,000 or more postings, the platform's display cap. Treat this one as an upper bound, because the keyword also matches conventional QA jobs that merely mention AI-assisted testing.

Two honest caveats. First, keyword counts are demand signals, not deduplicated headcounts; one opening can match several keywords. Second, this is our baseline pull. The quarter-on-quarter deltas this series promises will appear from the next scheduled count onward, measured against the 6 August 2026 figures above.

Who is hiring, and through which doors

Read the employer names, not just the totals. Sutherland is advertising data annotation analysts. TaskUs wants video data annotators. Innodata, a data-for-AI firm with a large Philippine operation, is posting AI content evaluators and LLM audio and video annotation specialists. Concentrix lists an AI operations manager. Accenture in the Philippines is hiring contact centre AI solution consultants. These are the same firms that hire voice agents by the thousand, which matters for anyone trying to move internally rather than start over.

The doors are not all the same height. Annotation and content evaluation sit lowest: several listings state that the work is browser-based labeling with no coding required. Conversation design and AI quality assurance sit in the middle, and they reward exactly what tenured agents have, which is deep familiarity with how customer conversations actually go wrong. Prompt-adjacent engineering roles sit highest and increasingly ask for scripting or LLM tooling experience alongside the writing skill.

What the roles pay, and why salary pages deserve suspicion

Pay data for these roles is thinner than posting data, and some of it is bad. One vendor rate card from staffing firm Second Talent, updated 18 December 2025, puts full-time data annotator pay at ₱18,000 to ₱32,000 monthly for juniors, ₱32,000 to ₱55,000 at mid level, and ₱85,000 to ₱120,000 or more for lead and QA specialist roles. That is a self-published marketing document from a firm that sells staffing, so treat it as indicative, not verified.

Now the cautionary example. Indeed Philippines reported an average prompt engineer salary of ₱898,876 per month as of 4 February 2026. That figure rests on three reported salaries and is almost certainly a mislabeled annual number. We cite it because you will meet figures like it everywhere: before trusting any salary page, find the sample size. Under ten reports, the number is noise.

The industry backdrop: growth and reshaping at once

The macro numbers explain why these roles are appearing inside BPOs rather than beside them. The industry closed 2025 with roughly 1.9 million workers and 40 billion dollars in export revenue, having added about 80,000 jobs during the year, and IBPAP targets nearly 1.97 million jobs and 42 billion dollars for 2026. At the same time, reporting by the Manila Times on a February 2025 IMF analysis notes that around one third of Philippine jobs are highly exposed to AI and that roughly 57,000 BPO workers, about 3 percent of the workforce, face immediate displacement risk. The same report states that 67 percent of Philippine BPO firms say they have already implemented AI.

The response is funded, which is worth knowing when you pick training. The industry association has committed at least 25 million dollars per year to workforce development, and Project UNLAD, a ₱740 million upskilling initiative with DICT and TESDA, is now running. IBPAP president Jack Madrid put the direction plainly in January 2026:

What got us here will not be enough to take us where we need to go next. Our focus in 2026 is to relentlessly upskill our workforce, embrace higher-value work and continue working closely with government, academe and investors.

How to use these numbers if you want in

Three practical moves follow from the data. First, target employers already posting, because internal transfer beats cold application; if you are at Sutherland, Concentrix, TaskUs, or Accenture, the AI roles above may be one job board away on your own intranet. Second, enter through the widest door you qualify for. Annotation and content evaluation postings ask for attention to detail and reliable internet before they ask for code. Third, build evidence, not just certificates. A prompt library you wrote, a labeled dataset you QA'd, or a chatbot flow you fixed will read better than a course badge alone. If you want a structured path from agent work into these roles, the AI Academy maps skills to each of the six roles counted here.

We will re-run these counts quarterly. The next report will show which of these roles grew, which stalled, and by how much, against the dated baseline above.