Mehrdad Yazdani

About Mehrdad Yazdani

I’m most interested in machine learning problems where the challenge is defining the right objective, designing the right evaluation, and understanding the trade-offs. My interests include research-oriented ML, especially preference learning, reward modeling, off-policy evaluation, optimization, retrieval, and deep learning systems.A lot of my work has involved building and improving real-world ML systems, but the part I enjoy most is the research side: designing methods, running thoughtful experiments, understanding failure modes, and developing approaches that generalize beyond a single product surface.

Experience

1. Machine Learning Engineer

   Meta

   Nov 2020 — Present · US

   Led ML work across WhatsApp integrity, GenAI, and Facebook recommendation/personalization surfaces- Worked on model evaluation, experimentation, and optimization, including reward modeling, preference learning, and retrieval-based methods- Improved production models through architecture exploration, hyperparameter analysis, data quality investigation, and failure-mode analysis- Built and extended tooling for model iteration, evaluation workflows, and ML development efficiency- Drove cross-functional work spanning modeling, infrastructure, experimentation, and launch readiness.

Education

• UC San Diego

  PhD, Intelligent Systems, Robotics, and Control.

  2008 — 2012

Skills

• Artificial Neural Networks
• Robotics
• Signal Processing
• Statistics
• Computer Vision
• Optimizations
• Numerical Analysis
• Algorithms
• Machine Learning
• Scientific Computing
• Matlab
• Control Theory
• Programming
• Optimization
• Python
• Latex
• Computational Neuroscience
• Data Analysis
• Digital Signal Processing
• Artificial Intelligence
• Image Processing
• Data Mining
• R