Assistant Professor, New York University (2022-present)
Graduate Courses
FRE-GY7773 – Machine Learning in Financial Engineering
In this course we will give an overview of several applications of machine learning to capital market forecasting and credit modeling, beginning with regressions, “shallow” layered machine learning models (e.g. Support Vector Machines, Random Forests), and ending with “deep” layered machine learning models (e.g. Long Short Term Memory Networks). Each model is discussed in detail as to what input variables and what architecture is used (rationale), how the model’s learning progress is evaluated and how machine learning scientists and capital market traders evaluate the model’s final performance so that by the end of the course, the students should be able to identify the main features of a machine learning model for capital market forecasting and to evaluate if it is likely to be useful and if it is structured efficiently in terms of inputs and outputs.
Topics:
1. Training and testing workflow: scaling, cross-validation pipelines. Gradient descent: mini-batch, stochastic.
2. Financial metrics: profitability and risk.
3. Financial feature engineering.
4. Models: multivariate regression, logistic regression, support vector machines, principal component analysis, decision trees, random forests, k-means, and hierarchical clustering, Gaussian mixtures, MLPs, LSTMs, and auto-encoder neural networks.
5. Applications: credit modeling, financial time-series forecasting, investment portfolio design, and spread trading, credit cycle regime identification.
VIP-GY5000 BD – Active Portfolio Management with Machine Learning and Time Series Forecasting
This project course is intended to explore and discover new methods to optimize portfolio allocation using artificial intelligence and machine learning techniques such as hidden Markov models, Kalman filters, random forests, genetic programming, deep learning, reinforcement learning, etc. This project is an opportunity to apply the ‘learning by doing’ methodology and prepares students for quant or research type positions.
VIP-GY5000 BE – Merger & Acquisition Outcome Prediction
This project course is intended to explore and discover new methods for merger and acquisition outcome prediction using artificial intelligence and machine learning techniques such as hidden Markov models, Kalman filters, random forests, genetic programming, deep learning, reinforcement learning, etc. This project is an opportunity to apply the ‘learning by doing’ methodology and prepares students for quant or research type positions.
Teaching Assistant, University of Toronto (2018-2022)
Graduate Courses
MIE1615 – Markov Decision Processes
This course covers the mathematical foundations of the Markov decision processes such as Banach fixed point theorem, and contraction mapping, as well as the standard solution methods such as the value iteration and the policy iteration. It also addresses structural properties of optimal policy in relation with mathematical characterization of the value function and introduces the simulation-based method and the reinforcement learning.
Topics:
1. Fundamental concept for the Markov decision processes
2. Structural properties of optimal policy
3. Standard methods: the value iteration and the policy iteration
4. Simulation-based optimization
5. Reinforcement learning
TA Mentor Program – Faculty of Applied Science and Engineering (FASE) TA Training course
The Teaching Assistants’ Training Program (TATP) is a peer-training program providing pedagogical support to the three campuses of the University of Toronto, through the Centre for Teaching Support & Innovation. TATP currently serves teaching assistants and graduate students.
Undergraduate Courses
MIE567H1 – Dynamic and Distributed Decision Making
This course is to provide fundamental concepts and mathematical frameworks for sequential decision making of a team of decision makers in the presence of uncertainty. Topics include Markov decision processes, reinforcement learning, theory of games, multi-agent reinforcement learning and decentralized Markov decision processes. The course is technical by nature and for advanced students with strong mathematical background and programming skills.
Topics:
1. Markov decision processes
2. Reinforcement learning
3. Stochastic games
4. Multi-agent reinforcement learning
5. Decentralized Markov decision processes
MIE364H1 – Methods for Quality Control and Improvement (Teaching Award Certificate)
In manufacturing and service industries alike, quality is viewed as an important strategic tool for increasing competitiveness. Continuous quality improvement is a key factor leading to a company’s success. With more emphasis on quality, the cost and the product cycle time are reduced and the communication between producer and customer is improved. The course focuses on the following topics: introduction to quality engineering, TQM, quality standards, supplier-producer relations and quality certification, costs of quality, statistical process control for long and short production runs, process capability analysis and acceptance sampling, quality certification, six sigma quality, quality improvement using designed experiments and an overview of the Taguchi Methods.
MIE367H1 – Cases in Operations Research
This course focuses on the integration of the results from earlier operations research courses and an assessment of the different methods with regard to typical applications. The course is taught using the case method. Students are expected to analyze cases based on real applications on their own, in small groups and during lecture sessions, and solve them using commercial software packages.