Responsibilities:

• Design and productionize agentic AI frameworks , including multi-agent coordination, planning, tool-use, and memory , that allow agents to maintain long-term context and execute complex tasks across the Dropbox ecosystem.
• Lead the end-to-end design of ML systems, from fine-tuning (SFT, RLAIF) and advanced prompting to inference optimization and production monitoring.
• Establish rigorous safety, alignment, and evaluation frameworks to ensure our autonomous systems are helpful, honest, and harmless.
• Collaborate across Product, Design, Infra, and Frontend teams to translate ambiguous user needs into concrete AI capabilities that move the needle for the business.
• Mentor junior engineers and serve as a core contributor to the broader Dropbox AI strategy, fostering a culture of technical excellence.

Requirements:

• BS, MS, or PhD in Computer Science, Mathematics, Statistics, or a related quantitative field (or equivalent work experience).
• 8+ years of software engineering experience, with at least 5+ years dedicated to building and deploying production-scale AI/ML systems.
• Professional experience in ML modeling for complex systems such as Search, Ranking, or Recommender Systems.
• Deep familiarity with LLM architectures and hands-on experience with ML libraries (e.g., PyTorch, JAX, or similar).
• Strong proficiency in Python (required) and experience with systems languages like Go or C/C++. You should be comfortable building the infrastructure that surrounds the model.
• Extensive experience working with large-scale distributed data systems and high-throughput production environments.
• Exceptional analytical skills and a 'bias to action' when navigating ambiguous technical challenges.

Preferred Qualifications:

• PhD with a focus on Deep Learning, NLP, or Reinforcement Learning (RLHF/RLAIF).
• Proven track record of taking AI products from concept to launch, either at a massive scale (millions of users) or by leading multiple 0 → 1 cycles in a fast-paced environment.
• Hands-on experience with autonomous agent frameworks, multi-step planning, tool-use (function calling), and advanced RAG.
• Experience with inference optimization, model distillation, or fine-tuning techniques to improve performance and cost-efficiency.

XML job scraping automation by YubHub