Felipe Godoy

I'm a Statistics & Data Science masters student at Stanford University's Department of Statistics. 

My main interests are:

• High-dimensionality regression modelling problems under evolving conditions 

• Nonparametric algorithms for statistical inference

• Mathematical theory for computational supervised learning (ranging from linear models to deep learning)

• Embeddings spaces & representation learning 

About me:

I completed my B.S. in Computer Science at Georgia Tech in 2021 (highest honors) with concentrations in Artificial Intelligence and Theory of Computation. I dedicated most of my advanced coursework to the Mathematical principals underlying Machine Learning and Artificial Intelligence: Linear Algebra, Applied Statistics, Algorithms, and Differential Equations. 

During my undergraduate studies I served as the Head TA for 'Intro to Artificial Intelligence', and a TA for 'Differential Equations' and 'Intro to Object Oriented Programming'. I also conducted research in the Center for Relativistic Astrophysics under Dr. Laura Cadonati and in partnership with LIGO (Laser Interferometer Gravitational-Wave Observatory), designing computational learning models to detect glitches in raw gravitational wave data.

In the Fall 2021, I started my M.S. in Statistics & Data Science at Stanford, where I have been fully funded by teaching assistantships in the School of Mathematics and the School of Computer Science. My graduate coursework has been dedicated to Statistical Inference, Stochastic Methods, Numerical Linear Algebra, Statistical Learning Theory, Scientific Computing, and Large-Scale Machine Learning. My graduation date is June 2023.

Towards my M.S. I conducted my independent research project to develop a nonparametric inferential algorithm which tracks the evolving behavior of ML models deployed to operate on shifting data streams, and quantify its evolving probability of outputting anomaly values.  

I'm currently a graduate student researcher under Dr. Andrew Ng in the Stanford ML Group, researching mislabeled data detection in large high-dimensionality image datasets; and in the Computational Neuroimaging Science Lab under Dr. Pohl, developing training techniques with representation learning components in neuroimaging models. I've also served as the Head TA for  'Design & Analysis of Algorithms', and as a TA for 'Modern Applications in Linear Algebra & Multivariable Calculus' and 'Theory of Computation'.

Formerly, I have held a Research internship at Facebook in the Forecasting Team, researching fast multivariate time series embedding algorithms for high-volume forecasting applications. And previously, also at Facebook, I interned in Data Science, working with optimizing ML models to reduce bias in content-enforcement automated decisions.

Interests:

My interests in Math & CS are varied, but they converge in Computational Statistics, Statistical Learning Theory, and Algorithm Design. Below is a short sample of my favorite topics and applications:

ML Theory: performance robustness, decaying performance, learning under distribution shifts, representation learning

Applied Statistics: high-dimensionality regression, nonparametric inference, anomaly detection

Algorithms: ranking algorithms, recommender algorithms, numerical algorithms, nonparametric algorithms

Linear Algebra: embedding spaces / latent spaces, dimensionality reduction

Key Moments: