
Editorial Prompt Engineer, ITP Luxury Group
- Dubai
- Permanent
- Full-time
- Build and maintain prompt libraries (headline packs, SEO/meta, outlines, interview preps, listicles, explainers, social captions, scripts, show notes) per brand voice.
- Create multi-step prompt chains for drafting → refinement → fact-check → style pass; version prompts, run A/B tests, document what works.
- Map end-to-end editorial workflows and insert AI at the right steps (research assistance, angle generation, text polishing, packaging), always with human-in-the-loop sign-off.
- Stand up repeatable automations (e.g., CMS → draft package → SEO → social variants) using off-the-shelf tools (no-code/low-code) and enterprise LLMs.
- Design “verify mode” prompts that force sourcing, quote extraction, claim lists, and line-by-line evidence checks; require links and provenance for all assertions.
- Create hallucination-reduction patterns (structured checklists, constrained schemas, refusal triggers).
- Build prompts for entity/keyword clustering, meta descriptions, URL slugs, image alt text, and headline/subhead variants tailored to each title.
- Provide platform-native social packs (IG/TikTok/YouTube/Threads) with hooks, lengths, and CTAs matched to channel behaviours.
- Prompt frameworks for video/podcast packaging: titles, descriptions, chapters, thumbnails copy, shownotes, and transcript clean-up.
- Generate multi-ratio copy variants and localised (EN/AR) versions with style checks.
- Run editors’ rooms/office hours, lunch-and-learns, and build concise playbooks and style-infused system prompts per brand; track adoption and outcomes.
- Coach teams on prompt hygiene, disclosure, and when not to use AI (sensitive topics, high-risk reporting).
- Embed newsroom AI principles: transparency, human oversight, labelling of AI-assisted outputs, and data/PII protection; align with AP/Reuters/BBC-style guidance.
- Set guardrails for branded work (church & state), copyright/citations, and content labelling for any automated assistance.
- Define quality rubrics (accuracy, originality, tone, fairness) and productivity metrics (turnaround time, accepted-as-final rate, edits per draft).
- Report lift in pageviews/engagement for AI-assisted packaging and measure subscriber/loyalty impact on premium brands.
- Compare models/tools (ChatGPT Enterprise, Claude, others) for task fit, safety and cost-performance; maintain a menu of approved use cases.
- Pilot retrieval-augmented workflows using brand archives (licensed content, past issues) to keep outputs on-brand and verifiable (e.g., “Ask FT”-style constrained assistants).
- Monitor public trust signals around AI in news; propose disclosure language and FAQ for audiences and clients.
- Liaise with legal/compliance on rights, data retention, and crawler policy posture.
- Create/update prompt kits per brand vertical (fashion, beauty, luxury, business, culture).
- Set up prompt testing boards (with examples, failures, and fixes).
- Pair-prompt with editors on live commissions; coach headline desks on fast, ethical AI use.
- Build macro-templates for newsletters, franchise pages, and social series.
- Maintain a changelog of prompt iterations, success rates, and guidelines.
- Draft AI use disclosures and contributor guidance for contracts.
- 6–10 years in editorial, copy, or audience roles with demonstrable LLM/prompting expertise; portfolio of before/after improvements (headlines, SEO, social, longform).
- Excellent copy craft; ability to encode tone/voice into prompts and templates.
- Comfort with structured prompts, JSON/YAML schemas, and light automation (Sheets/Notion/Zapier/Make); familiarity with RAG basics a plus.
- Strong grasp of newsroom standards (sourcing, fairness, labelling) and AI governance concerns.
- Data-literate: can define KPIs, analyse outcomes, and recommend next steps.
- Bilingual Arabic/English strongly preferred.
- Cycle-time reduction from commission to publish for AI-assisted tasks.
- Accepted-as-final rate for AI-assisted drafts/packaging.
- Accuracy/standards compliance (zero critical incidents; declining correction rate).
- SEO/Social uplift on AI-assisted packages (CTR, rank movement, engagement).
- Adoption & satisfaction: % of teams using approved prompt packs; training NPS.
- Cost-efficiency: tokens/runtime vs. time saved and output quality.