About this role
Summary: Surveys and evaluates time series foundation models against CFM's forecasting needs, then researches and implements techniques to adapt them. Tests these models out of sample and helps plan their integration into firm workloads.
ABOUT CFM
Founded in 1991, we are a global quantitative and systematic asset management firm applying a scientific approach to finance to develop alternative investment strategies that create value for our clients.
We value innovation, dedication, collaboration, and the ability to make an impact. Together, we create a stimulating environment for talented and passionate experts in research, technology, and business to explore new ideas and challenge existing assumptions.
Your Role
The Data-Human-Insights team works across a wide variety of structured and unstructured datasets, from time series to graphs, assessing their predictive power and preparing them for modeling. As data scientists, we work closely with the Research team to extract valuable insights from the data.
Accurate and efficient time series forecasting is central to CFM's work, and it sits at one of the most important intersections between the expertise and interests of researchers and data scientists. The literature is large and growing, spanning econometrics, operations research, traditional machine learning, deep learning and generative AI. In recent years, time series foundation models have become a performant and popular option for a wide variety of forecasting tasks.
The goal of this internship is to survey and evaluate existing time series foundation models against CFM's needs, and to identify where they can improve on the forecasting methods we use today. Building on this evaluation, you will research, propose and implement techniques to adapt these models to our purposes, and help plan and develop their integration into different workloads across the firm.
You will be expected to understand, implement, evaluate and operationalize contemporary research on time series foundation models, and to test it as rigorously out of sample as any other model we use
Key Responsibilities
- Survey the academic and industry literature on time series foundation models and identify the most relevant approaches.
- Benchmark existing models against current forecasting methods, including classical approaches (ARIMA, GARCH, state space models).
- Propose and implement techniques to improve these models for CFM's use cases (e.g. fine-tuning with LoRA, quantization, architectural adaptations).
- Evaluate models rigorously out of sample, avoiding look-ahead bias and data leakage.
- Aggregate and join heterogeneous data sources and process huge datasets efficiently.
- Help plan and develop the integration of selected models into different workloads.
- Write clear documentation and reports on your methodology, results and limitations, and present them to the team and, occasionally, to researchers.
Your Skills
You are in final year of a Master's degree or engineering school program in data science, statistics, applied mathematics, computer science or a related field.
You are proficient in Python, pandas, scikit-learn, PyTorch and Matplotlib.
You have experience manipulating and modeling time series, with a solid grounding in classical approaches such as ARIMA, GARCH and state space models.
You have experience with modern deep learning and generative AI architectures and techniques, such as transformers, attention mechanisms, positional encodings, fine-tuning with LoRA, and quantization.
You understand how to evaluate forecasting models out of sample, including how to avoid look-ahead bias and data leakage.
You can aggregate and join heterogeneous data sources and manipulate huge datasets.
You can conduct independent literature reviews, propose new ideas, and synthesize results, challenges and proposals.
You are proactive, with a drive to experiment and propose original ideas.
You communicate results and limitations clearly and concisely.
You are fluent in English.
You are available for a six-month internship.
EQUAL OPPORTUNITIES STATEMENT
We are continuously striving to be an equal opportunity employer and we prohibit any discrimination based on sex, disability, origin, sexual orientation, gender identity, age, race, or religion. We believe that our diversity, breadth of experience, and multiple points of view are among the leading factors in our success.
CFM is a signatory of the [Women Empowerment Principles](https://www.weps.org/about).
FOLLOW US
Follow us on [Twitter](https://twitter.com/cfm_am?lang=fr) or [LinkedIn](https://www.linkedin.com/company/419383/) or visit our [website](https://www.cfm.com/) to find out more about CFM.