AI Video Ad Creative Specialist: Fashion Brand (Higgsfield / Kling / Veo / Seedance)

AI Video Ad Creative Specialist — Fashion Brand (Higgsfield / Kling / Veo) Hi! I run a fashion e-commerce brand selling fashion to customers in US and UK. We scale through Facebook and Instagram ads, which means creative is the whole game, we need a constant supply of new angles, hooks, and formats to test. I'm looking for one person to own AI ad creative production end to end: concept, script, generation, voiceover, edit, export. You'll be building ads with AI generation tools and cutting them into direct-response creatives that sell. This is a long-term role and the output is measured in performance, not pretty edits. What you'll produce Short-form (this is the bulk of the work, under 60 seconds) Hook-led ad creatives, typically 15–45 seconds 3–5 hook variations for every concept that shows promise 10–20 second retargeting cuts for warm audiences Static frames and thumbnails pulled from the same shoots Long-form (a smaller number each month, 60 seconds to 3 minutes) VSL-style ads with a full script arc: hook ? problem ? agitation ? product reveal ? proof ? offer ? CTA Paced so the viewer stays past the 30-second drop-off, with visual changes every 2–4 seconds Every deliverable ships in 9:16 and 4:5 (1:1 when requested), correct safe zones, burned-in captions Clean file naming and organized folders in Google drive What the work actually involves Research angles and formats in the Meta Ad Library and [WinningHunter / Foreplay] — we study competitor structure and angles, we never reuse their footage, music, voiceover, or claims Adapt scripts using direct-response frameworks (PAS, hook–story–offer, problem–solution, testimonial) Generate video with AI models and get consistent results — same model/character, same wardrobe, same lighting across every shot in an ad Produce AI voiceover in our locked brand voice and sync it cleanly to the visuals Assemble in [CapCut / Premiere Pro]: captions, b-roll, real product footage, music, sound design, pacing Batch-produce hook variants so we can test cheaply Take performance feedback (hook rate, hold rate, CTR, ROAS) and cut new versions from what's working Maintain our creative asset library, prompt library, and naming conventions so nothing gets rebuilt twice Required — please only apply if you have all of these A portfolio of AI-generated video ads you personally made, with the tool and model named for each piece Hands-on experience with Higgsfield (this is our main pipeline) — or deep experience with an equivalent workflow plus a fast, demonstrated ability to learn it Working knowledge of current video models: Kling 3.0, Veo 3.1, Seedance, Sora 2, Runway, Wan — you know which model to reach for and why AI voice generation (ElevenLabs or similar) and clean audio sync Strong editing in CapCut, Premiere Pro, or After Effects You understand direct-response advertising: what a hook is, why the first 2 seconds d ---------- everything, and what a hold-rate drop-off means Good visual taste for fashion — you can tell when AI output looks fake, off-brand, or cheap, and you fix it instead of shipping it Good written English and a reliable machine and internet (rendering and uploading all day) Bonus points for Apparel, beauty, or accessories brand experience AI UGC avatar tools (Arcads, Creatify, HeyGen) Image models for stills and shot references (Midjourney, Flux, Nano Banana) Motion graphics, price/offer overlays, comparison frames