Prompt Engineer | I Help U.S. Brands Create High-Converting AI Ads with ChatGPT & Midjourney
Volley is Hiring: Prompt Engineer / AI Workflow Specialist (Contract, Remote – U.S. Time Zone Overlap)
Duration: ~8 Weeks (Part-Time, with Extension Possible)
Pay: $80–$110/hr USD
Start: ASAP
Volley is the leading developer of voice-controlled games for Amazon Alexa, Roku, and Fire TV. From trivia hits like Song Quiz to immersive experiences like Jeopardy! and 20 Questions, we bring engaging, voice-first games to millions of households. As we continue scaling, our Growth and Design teams are investing in generative AI to streamline asset production and unlock creative speed at scale.
We’re looking for a freelance Prompt Engineer / AI Creative Technologist to help us build a repeatable, documented pipeline that transforms creative briefs into production-ready images and motion assets using state-of-the-art generative AI tools. If you know your way around Midjourney, Stable Diffusion, Runway Gen-2, and scripting automation workflows—and can translate brand guidelines into generative templates—we’d love to work with you.
💡 The Mission
Your goal is to design and deliver a modular, scalable, and repeatable AI-powered content pipeline for our design team. You’ll help us move from idea to asset in hours—not days—by building smart, branded prompt systems and reusable motion generation workflows.
You’ll work closely with designers, marketers, and creative leads to ensure the output is visually consistent, on-brand, and ready for CTV (Connected TV) specs like Roku and Fire TV. You’ll also help document and train our team on how to use these AI tools independently.
🎯 Project Goals & Responsibilities
You’ll be responsible for the following high-impact deliverables:
✅ Prompt Library & Brand Replication
- Build a prompt library that accurately reproduces Volley’s visual brand styles using Midjourney or Stable Diffusion.
- Tune prompts to produce clean, consistent, and editable outputs based on creative briefs.
- Use seeds, custom LoRA files, or checkpoints to lock in stylistic fidelity and reproducibility.
✅ Motion Generation with Runway or Pika
- Create Runway Gen-2 workflows that can convert keyframes or image prompts into short-form motion clips.
- Build reusable templates for 15-second hero videos, animated backgrounds, and looping screensaver content.
✅ Automation & Output Prep
- Script simple automation flows in Python or JavaScript to batch-render outputs, apply safe zones, and resize assets according to platform specs (Roku, Fire TV).
- Integrate export presets into design tools like Figma, or generate plug-and-play overlays for designers.
✅ Internal Enablement
- Ship a Notion-based guide for internal designers, outlining how to use the system, tweak prompts, and regenerate assets quickly.
- Train the design team to be semi-autonomous in running and modifying the workflows you build.
- Maintain organized files of seeds, prompt history, checkpoints, style models, etc., for easy handoff.
✅ Optimization & Feedback Loops
- Monitor and improve output quality and token costs for tools like Midjourney, Stable Diffusion, and Runway.
- Optimize for cost, speed, and repeatability.
- Incorporate designer feedback and performance data to improve templates and prompt libraries over time.
📆 Timeline & Milestones
This is an 8-week part-time engagement, front-loaded in the first 2–3 weeks to build the initial systems. Approx. 5–8 hours per week, with flexibility for collaboration.
- Week 1: Review current assets, audit brand references, and deliver first visual prompt proof.
- Week 2: Finalize version 1 of the prompt library and document in Notion.
- Weeks 3–4: Build and test Runway/Pika templates for hero videos and background loops.
- Weeks 5–6: Automate asset prep workflows, define export presets, and link to Figma component library.
- Week 7+: Monitor results, gather feedback, and iterate based on creative team needs and ad performance data.
✅ Must-Have Qualifications
To be successful in this role, you should have:
- 2+ years of hands-on experience building real-world generative pipelines for design or marketing teams.
- Deep understanding of Midjourney parameters, Stable Diffusion checkpoints, and tools like ComfyUI.
- Strong working knowledge of Runway Gen-2, or similar tools (Pika, Sora, etc.) for motion generation.
- Ability to write clean SOPs, guides, and playbooks for non-technical creatives.
- Comfort writing scripts in Python or JavaScript for automation, batch processing, or resizing assets.
- Availability to overlap at least 2 hours with U.S. Pacific Time for live feedback and check-ins.
💡 Nice-to-Haves (Bonus Skills)
- Experience creating CTV-specific assets (Fire TV, Roku, Apple TV).
- Familiarity with Photoshop Generative Fill or Adobe Firefly.
- Experience fine-tuning or training LoRA models for custom brand styles.
- Background in color science, branding, or cross-language campaign styling.
🧰 Tools You’ll Use
- AI Image: Midjourney, Stable Diffusion, ComfyUI, ControlNet
- Motion AI: Runway Gen-2, Pika
- Design Tools: Figma, Photoshop (optional), Notion
- Automation: Python, JavaScript, shell scripts
- Communication: Slack, Loom, Linear
🔁 Workflow & Communication Style
- Kickoff Call: Align on goals, review brand visual identity.
- First Deliverable: Style prompt proof due in 48 hours.
- Revisions: Async feedback via Slack + Loom. Quick 24-hour turnaround.
- Weekly Sync: Check-in to review creative output, designer feedback, and refinements.
- Final Handoff: Deliver prompt library, workflow scripts, documentation, and video walkthroughs.
🚀 Why Work with Volley?
You’ll be joining a team that’s serious about using AI to accelerate creative work, not replace creativity. We value fast iteration, clear feedback, and building tools that non-technical designers can actually use. If you love making generative tools feel like magic — and you’re not afraid to get hands-on with prompts, scripts, and visuals — we want to hear from you.
📩 To Apply:
Send your portfolio or sample prompts, along with a brief note on your experience building AI creative workflows.