Andreas P. Mentzelopoulos

Hi, I’m Andreas — a Machine Learning Scientist at Wayfair!

As part of the Measurement & Attribution team, I develop multi-touch attribution models that quantify the impact of marketing campaigns via explainable deep learning (Shapley-based frameworks). My work helps guide the allocation of over $1B in annual marketing spend by providing transparent, data-driven insight into what truly drives customer engagement.

I am the founder and lead researcher of LOBSTgER, a project that develops latent diffusion models to generate and enhance high-resolution underwater imagery. In collaboration with photographer Keith Ellenbogen, LOBSTgER combines AI and ocean photography to raise awareness of marine environments while advancing generative modeling research.

I recently graduated with my PhD from MIT, where my research centered on deep learning, particularly generative modeling, and time-series forecasting. Specifically, I developed the Firstling Digital Twin, pioneering the use of transformers (VIVformer) and diffusion models (Virtual Towing Tank, publication under development) to forecast and simulate vortex-induced vibrations of flexible structures.

Beyond engineering, I have designed and implemented deep learning–based statistical arbitrage strategies for public equities, back-tested in Southeast Asian markets during my Wall Street internship with CITIC Securities, CLSA.

Originally from Athens, Greece, I studied at the University of Michigan before joining MIT. Outside of research, I am a violinist (past 5 years with the MIT Symphony Orchestra) and enjoy concerts, musical theater, and time with friends and family.