Career Profile
Dr. Nordlund is a science leader with over 15 years experience in applied machine learning and economic modeling. He research spans spillover effect modeling, long-term effect modeling, debiased machine learning, causal deep learning, synthetic counterfactual methods, experimental design, and applied natural language processing. Dr. Nordlund lives in Redmond, Washington with his wife, daughter, and two dogs.
Experiences
- Set long-term business vision for cross-functional projects that span product, science, and engineering.
- Grew the causal science team to double in size to address accelerating scope and breadth of impact to Amazon Worldwide Stores.
- Created $100M+ annual value from designing, building, and launching a new A/B testing platform for pricing algorithms.
- Worked with business partners from multiple organizations inside the company to deliver 150+ statistical models per year to drive launch decisions for machine learning systems.
- Set and led the research initiatives for machine learning models that serve promotions sourcing strategies for both day-to-day operations and large events (e.g. Prime Day).
- Invented a data compression and sampling algorithm to build causal forecasting models on top of large time-series data; presented at multiple internal machine learning conferences.
- Independently built and helped productionalized a causal, de-biased machine learning model to assess intertemporal effects of promotional pricing for worldwide promotions to drive improved sourcing of third-party deals.
- Developed the science for a multi-org initiative to optimize the pricing CX in search results.
- Oversaw the worldwide recruiting of Ph.D. students and advised the development of their research projects.
- Managed personnel assignments between graduate research assistants and university faculty.
- Designed, built, and administered a local compute server for JupyterHub and RStudio to replace existing courses managed on AWS/Kubernetes.
- Researched nonexperimental causal effects with a focus on neighborhood and edge-based spillover effects. Constructed network-valued, graph models on AWS high-RAM servers. Performed topic modeling, spectral clustering, and NLP tasks on high volumes of text data.