Staff Applied Scientist - AI Guardrails

Generative AI is redefining creativity. Ensuring that these systems are safe, controllable, and respectful of intellectual property is one of the most important open challenges in deploying generative models responsibly at scale. 

The Adobe Firefly Applied Science & Machine Learning team is building a unified guardrails platform that enables safe training and deployment of generative models across image, video, and audio. This includes large-scale automated data moderation systems, multimodal detection and safety mechanisms, and learning based on input that continuously improves system robustness.  

We are seeking a Staff Applied Scientist to help architect and scale this platform while leading key applied ML initiatives. This role sits at the intersection of multimodal machine learning, large-scale ML systems, and production AI safety infrastructure. The ideal candidate will drive high-impact technical initiatives while defining the architectural patterns that enable guardrail systems to scale across models, products, and enterprise use cases. 

Research Areas You Will Drive 

Guardrails Platform Architecture 

• Architect end-to-end systems that integrate data pipelines, detection, evaluation, and model control mechanisms. 

• Optimize systems for latency, throughput, and cost efficiency under real-world constraints. 

• Define reusable patterns that generalize across products and enterprise deployments. 

• Identify and address system-level bottlenecks to ensure scalable and reliable deployment. 

Safety-Aware Data Systems 

• Develop strategies to identify and mitigate intellectual property and trust & safety risks in large-scale datasets. 

• Advance methods to improve coverage and robustness of safety systems, particularly across long-tail and evolving concept spaces. 

• Drive system-level thinking around efficiency, scalability, and reliability of data-centric safety mechanisms. 

• Partner with modeling teams to ensure data quality and safety signals effectively translate into improved model behavior. 

What You’ll Do 

• Lead the architectural development of scalable, data-centric safety and guardrail systems across multimodal ML platforms. 

• Drive applied ML initiatives spanning data-centric safety, detection, and feedback-driven system improvement. 

• Translate research advances into production-ready systems, balancing model quality, scalability, and efficiency. 

• Act as a technical lead and mentor, guiding applied scientists and engineers while setting high standards for system design and execution. 

• Shape the long-term technical strategy for building robust, scalable, and responsible generative AI systems at Adobe. 

What You’ll Need to Succeed 

Research & Technical Depth 

• PhD or MS in Computer Science, Machine Learning, AI, or a related field. 

• 8+ years of experience building and deploying machine learning systems in production environments, with demonstrated end-to-end ownership. 

• Strong expertise in multimodal machine learning, including vision-language and generative models. 

• Experience working with large-scale datasets and applying data-centric approaches to improve system performance and robustness. 

• Proficiency in modern ML development ecosystems (e.g., Python, PyTorch), with the ability to translate research into scalable implementations. 

ML Systems Architecture 

• Experience designing scalable ML systems under real-world production constraints. 

• Strong intuition for trade-offs between model quality, latency, and infrastructure cost. 

• Experience integrating ML systems into large-scale product environments. 

Scientific & Systems Rigor 

• Strong experimental design and evaluation skills. 

• Experience analyzing complex system-level failure modes. 

• Ability to operate in open-ended problem spaces and identify high-leverage opportunities. 

AI-Accelerated Development 

• Experience using AI-assisted development tools to accelerate experimentation and system design. 

• Ability to combine rapid iteration with production-quality implementation. 

Collaborative Approach 

• Experience working in multi-functional research-to-product environments. 

• Strong communication skills with the ability to influence technical direction across teams. 

Preferred Qualifications 

• Experience with trust & safety systems, content moderation, or AI safety infrastructure. 

• Background in large-scale multimodal data processing or dataset curation. 

• Experience deploying ML systems serving large user bases. 

• Research contributions in machine learning systems, multimodal AI, or responsible AI. 

#FireflyGenAI