Keji Li Resume 2023-10-13

Areas of Expertise

quantitative programming and DevOps

End-to-end data pipeline development from data acquisition to model productionization.
Development of data visualization and trade automation tools for portfolio managers and traders.
Experience with Airflow, ActiveBatch, Github Actions, Gitlab CI/CD, Docker, and AWS S3/ESC.

statistics

Proficient in Structural Equation Models (SEM), Hidden Markov Models (HMM), and basic Partial Differential Equation (PDE).
Image processing including manual feature detection and CNN-based object detection/motion tracking.
Experience with Recurrent Neural Networks (RNN) and Variational Autoencoders (VAE).

programming languages and major libraries

Python: Pandas Polars Airflow Sklearn PySpark OpenCV XGBoost PyTorch SQLAlchemy
Rust: ndarray pyo3
C/C++: Boost Qt4 Qt5 OpenGL-ES-2.0
SQL, Spark, R: lavaan nlme

financial investment research

Factor research and model estimation for commodity models.

hardware development

Independent development of animal experiment rigs for motor learning experiments using: Digital circuit design. 3D modeling with AutoCAD/FreeCAD/Blender for 3D printing/laser cutting/assembly/rendering.

neuroscience research

Light, confocal and electron microscopy; 2-photon and 3-photon Ca²⁺ and Cl^- imaging of awake animals
3D reconstruction of brain areas and neural projections
Optical imaging of BOLD or Ca²⁺ signal

Professional Experience

Acadian Asset ManagementBoston, MA

Quantitative Developer

Aug 2020-Oct 2023

Led upgrade to ~300k LOC codebase safely from Python 2 to Python 3, then from Pandas 0.25 to 2.0. Improved ad-hoc integration test framework and trained team members for the upgrade.
Developed and productionized commodity models through close collaboration with portfolio managers.
Established a robust data pipeline for model estimation, including data retrieval, parsing, cleaning, transformation, and aggregation.
Turned previously loss-making models into profitable ones.
Contributed to the development of transaction cost integration in models across equity, bonds, and foreign exchange sectors.
Participated in the migration of pipeline orchestration from ActiveBatch to Airflow.
Automated and enhanced the trade generation system for variance swaps, reducing manual inventions from PM’s and traders.

Massachusetts Institute of TechnologyCambridge, MA

Post-doctoral fellow

Aug 2015-Aug 2020

Collaborated in neuroscience projects, analyzing neuronal encoding with vector regression (SVR), RNN and particle filtering; evaluated with a continuous estimate of mutual information. Developed behavioral task rigs integrated with 2p microscopy.
Directed a team in a translational project for a novel drug to treat Rett Syndrome, leading pre-clinical tests and statistical analysis using mixed linear models and non-parametric tests.
Created an automated pipeline for two-photon imaging data processing using both conventional and deep learning-based object detection/segmentation techniques.

Education

Vanderbilt UniversityNashville, TN

PhD in Psychology, neuroscience focus/quantitative methods secondary focus

Aug 2015

Fudan UniversityShanghai

BS in Life Sciences

Jul 2009

Honors and Awards

Rettsyndrome.org foundation grant #3213: mentored training fellowship

Role: Principle investigator

Synaptic mechanisms and novel therapeutic strategies for Rett Syndrome

2016-2018

The JPB foundation: JFDP post-doctoral Fellowship program

Role: Post-doctoral fellow

Synaptic mechanisms and novel therapeutic strategies for Rett Syndrome

2015-2016

Victor and William Fung foundation

Fung scholar fellowship

2006-2007

Ministry of education, China

People’s scholarship, second class

2005-2008

Keji Li

Data Scientist | Quantitave Programmer

Summary