NewsBreak
AI Engineer, Agentic Ad Creative (Multimodal)
$120K–$220K
Mountain View, California, United States · Full-time · Posted Jun 24, 2026
About the job
Own an end-to-end agentic ad-creative pipeline that generates, evaluates, and improves multimodal ads using performance feedback, while managing quality, policy compliance, and cost.
What you'll do
- Extend the CPI-feedback creative loop to generate and improve images, video, and copy at scale.
- Build video-ad generation, model-routing, and policy-checking pipelines; train and deploy brand-style LoRAs or checkpoints.
- Ship event-triggered, platform-specific ad variants and use A/B testing and performance data to improve the pipeline.
What they ask for
- Hands-on experience debugging diffusion and video models, training a LoRA, optimizing inference, and building ComfyUI workflows.
- Public artifacts demonstrating generative-AI engineering work.
- Ability to assess model strengths, limitations, and cost tradeoffs, and to ship cost-aware production pipelines.
- Familiarity with Meta Advertising Standards and TikTok branded-content rules.
Nice to have
- Pretrained or significantly advanced a VLM; or released an open-source ComfyUI node, diffusers extension, or Replicate model with traction.
- Scaled serverless inference on fal, Replicate, or Modal; or shipped an end-to-end video product.
Pay
$120,000 — $220,000 USD (base salary)
Tools
Flux, SDXL, Veo 3, Sora, Runway Gen-3/4, Kling, Hailuo, ComfyUI
AI focus
Build and operate agentic multimodal ad generation and optimization, including video creation, model routing, policy review using VLMs, and brand-style model training.
Model-written summary · the original posting is authoritative