Staff Applied AI Engineer

Arcade is building the world's first AI physical product creation platform, enabling users to design, purchase, and sell custom products using generative AI. They are seeking a Staff Applied AI Engineer to architect ML systems that support their generative platform, focusing on production reliability, quality evaluation, and system-level improvements.

Responsibilities:

• Architect the ML systems behind Arcade's generative platform — design the pipelines, training infrastructure, and serving systems that let diffusion models, LLMs, and emerging generative architectures run reliably and efficiently at production scale
• Find and build the high-leverage systems work — identify the architectural changes that scale our ML operations and meaningfully improve model output and quality, and lead them from design through production
• Set the evaluation and quality bar — design measurement systems (not one-off benchmarks) that real engineering decisions are made against, so model and pipeline quality is defensible with data
• Train, fine-tune, and deploy models — diffusion / text-to-image models and LLM-based applications, including advanced prompt engineering, fine-tuning, retrieval, and multi-component workflows
• Own production reliability — deploy, monitor, and maintain models in cloud-based production environments, ensuring scalability, latency, and cost are engineered intentionally
• Build the data foundation — design systems to collect, clean, and analyze large-scale datasets that improve model performance and reliability over time
• Bring state-of-the-art practice in-house — stay current with developments in generative AI, evaluate what's worth adopting, and translate the best of it into our pipelines
• Set technical direction and raise the bar — establish the patterns and standards the rest of the AI team inherits, and mentor engineers as the team scales
• Communicate clearly across audiences — make technical tradeoffs legible to both technical and non-technical partners across engineering, product, and design