Week 1- Introduction and acclimation
This week we begin by welcoming students to Pagaya and setting up the technical foundations needed throughout the fellowship. During this week you will get to know your mentors and attend introductory lectures from Pagaya employees who will talk about their work as researchers and data scientists at the company. In parallel, we’ll set up your programming environment for the coming weeks and offer a refresher course in OOP/Python.
Week 2- Hit the ground coding
This week we will delve into more advanced topics in Python with direct relevance to machine learning. These include familiarizing students with the key packages used for scientific research and handling big data in Python. In addition, we will continue our introduction to data science and machine learning at Pagaya and make our first attempts at writing research-oriented code while gaining a broad understanding of the research process at Pagaya. At the end of each day, you will have extended hands-on practice sessions in which you will implement what you’ve learned in Python with assistance from our mentors.
Week 3- First steps in ML
Once you’ve gained strong foundations in coding, we’ll dive right into the core concepts in supervised learning — a general term for ML techniques aimed at learning from labeled data. We will focus on the gradient descent algorithm, which will accompany us throughout the fellowship as we advance in optimizing more complex machine learning algorithms. In addition, you will also acquire practical tools to evaluate and improve your models.
Week 4- Statistical modeling 101
Working as a data scientist at Pagaya requires strong foundations in statistical modeling and hypothesis testing. During this week, we will focus on the key statistical tools we will need throughout the project, including sample analysis, statistical tests, likelihood estimators, and inferential statistics. We will also further study the connections between statistics and machine learning and apply the knowledge we gain this week in hands-on exercises.
Week 5- Exploratory data analysis
As the weeks progress, our activities become more and more implementation oriented as we increase our focus on the completion of your own industry-ready research project by the end of the fellowship. In this week, you will split into research teams, each with a specific research mission, which you will complete under the guidance of our mentors. You will gain access to your available data and begin processing it (e.g. detecting sampling noise and adverse effects, and working with large amounts of data). In addition, this week’s exercise sessions are aimed at further exploring some of the more advanced technical tools necessary for your future work as a data scientist in general, and for your projects in particular.
Week 6- Advancing in Machine learning
This week we’ll study the key machine learning algorithms in use today, from basic linear and logistic regression through decision trees and all the way to tree ensembles and boosting. Much of this week is also invested in familiarizing ourselves with some of the relevant machine learning packages available (e.g. SciPy, Scikit-learn and Xgboost) and with tools for evaluating and further attenuating and improving your models as you progress through your project.
Week 7- Advanced statistical methods
During this week we’ll continue with some of the more advanced statistical tools necessary for your future careers as data scientists. These include advanced modelling tools, classification and dimension reduction, resampling, and bootstrapping. As always, study of these techniques will be accompanied by intense hands-on sessions where they will be applied to various questions arising from your own projects.
Week 8- Unsupervised learning and classification
Up until now, we have mainly focused on supervised learning techniques, which are noted to make some revolutionary advancements in all facets of our lives. However, these techniques necessarily assume that our data is labeled, which is often not a plausible assumption to make.
During this week we’ll explore the kinds of machine learning algorithms that allow us to work with unlabeled data.
Week 9- Additional AI algorithms
Machine learning algorithms, while being incredibly foundational and important for the work of a data scientist, are only part of the diverse scope of artificial intelligence abilities applied throughout the data science world. This week we’ll take a closer look at some of the important evolutions of machine learning and their applications, including image processing, computer vision and natural language processing.
Week 10- So, how does it all fit together?
Before transitioning to the final stage of our fellowship, we’ll take a look back at what we’ve learned in the previous weeks and contemplate some future research question which may arise. This week is filled to the brim with guest lectures and hands-on sessions from Pagaya’s research staff, who will introduce you to some of the challenges and questions arising frequently during their work as data scientists and allow you to experience some of these yourselves before setting off on your independent research.
Week 11--15 - Final projects and advanced seminars
The last phase of the fellowship will differ from the previous weeks in that most of your time will be invested in hands-on practice sessions in which you will work on your independent projects under the guidance of your mentors. In parallel, we will continue with advanced lectures and seminars on more specialized topics in data science and machine learning to further your knowledge and understanding of these subjects. The end goal is for each student to be able to state in confidence that she has developed an end-to-end solution applying machine learning to an interesting research question. You will present your projects during the closing event of our fellowship.