Principal Research Scientist

A little about us

The Yahoo DSP Research team sits at the core of our programmatic ad platform, developing the mathematical models and algorithms that govern billions of dollars in real-time ad spend daily. We blend control theory, machine learning, and economics with modern AI-driven engineering practices to build world-class marketplace optimization systems.

As a Principal Research Scientist (IC6) on the Yahoo DSP Optimization team, you will serve as a primary technical authority driving the next-generation architecture for budget control, pacing, and bidding algorithms. In this role, you will lead the evolution of our high-scale optimization stack by integrating agentic AI practices—shifting traditional research-to-production pipelines into highly automated, AI-augmented development loops.

• Lead the research, architectural design, and production deployment of large-scale algorithms for budget pacing, bid optimization, and marketplace supply/demand balancing.
• Architect and spearhead the implementation of agentic AI workflows—utilizing AI coding agents, automated experiment orchestration, and agent-driven validation to dramatically accelerate research-to-production iteration velocity.
• Provide technical strategy and long-term vision for the DSP optimization ecosystem in close alignment with business and engineering leadership.
• Apply control theory, convex/stochastic optimization, and machine learning to solve complex, real-time marketplace dynamics under signal uncertainty.
• Direct AI-assisted code refactoring, automated testing suite generation, and LLM-driven research synthesis using modern tools (e.g., Cursor, Git Hub Copilot, Gemini, Claude, ChatGPT) to elevate team engineering standards.
• Own multi-quarter research initiatives from mathematical formulation through production rollout, establishing rigorous metrics and evaluation frameworks.
• Partner closely with engineering and product management teams to ensure algorithmic designs execute seamlessly at scale.
• Mentor senior scientists and engineers, setting technical excellence standards and fostering an AI-forward research culture.
• Communicate technical strategy, experimental results, and architectural choices clearly to executive and cross-functional partners.

• Ph.D. degree in Computer Science, Electrical Engineering, Operations Research, Statistics, Mathematics, or a related quantitative field.
• 8+ years of hands‑on experience designing and deploying large-scale optimization, control systems, or machine learning models in production environments.
• Deep mathematical foundation in probability, convex/stochastic optimization, and control theory (e.g., feedback control, model predictive control).
• Demonstrated expertise in production programming languages (Python, Java, C++, SQL, or Spark) with experience in low-latency or distributed data architectures.
• Active hands‑on experience with AI pair-programming and developer automation tools (e.g., Git Hub Copilot, Cursor, Claude Code, Codex, agentic orchestration frameworks).
• Proven capability to evaluate, refine, and validate AI‑generated research artifacts and code, ensuring strict safety, accuracy, and performance standards.
• Demonstrated mindset of continuous experimentation, leveraging AI tools to automate repetitive engineering and diagnostic tasks.
• Strong communication skills with a proven track record of translating complex mathematical concepts into clear executive strategy.

• Prior research or production experience in computational advertising, real-time bidding (RTB) auctions, or dual-sided marketplace design.
• Portfolio of peer-reviewed publications or patents in computational advertising, control theory, or applied optimization.
• Experience leading system‑level re-architectures utilizing agentic workflows or automated research pipelines.
• Familiarity with high-throughput, low-latency distributed computing frameworks.

• Deliver measurable efficiency gains in budget delivery and bid efficiency by deploying refreshed control models built via AI-augmented development loops.
• Establish standardized agentic development…