SETH W. EGGER, PhD Durham, NC 916-***-**** *********@*****.***
/www.linkedin.com/in/seth-egger
/github.com/swegger
Epic Games Seth W. Egger 7.17.25
SUMMARY
Seasoned Machine Learning and Artificial Intelligence Scientist with over 15 years of experience applying advanced AI/ML techniques to complex data domains. Expert interrogator of diverse, high- dimensional datasets using cutting-edge machine learning, Bayesian, and recurrent neural network (RNN) approaches. Recognized as a collaborative leader and mentor, with a strong track record in both academic research and cross-functional team environments. Accomplished technical communicator with a robust publication history and the ability to clearly convey insights to varied audiences. WORK EXPERIENCE
Senior Research Associate, Duke University 2019 to 2025 AI: Neural Network solutions for executing intelligent policies in real-time
• Engineered encoding and decoding models of image processing o Innovated decoding architectures to reduce estimation error by 25% (Python) o Applied theory to candidate architectures to derive experimentally testable hypotheses o Developed a framework to dissociate signal from noise in hidden layer activity
• Modeling intelligent policy execution through RNN engineering o Inferred connectivity from sensory encoding layer via reduced rank regression (Python) o Engineered a RNN to compute a policy from encoding layer activations, improving model predictive power by 39% and 20% for behavioral and neural data, respectively (Python) o Trained network with custom input class to emulate real-world data (TensorFlow) Machine Learning: inference strategies for high dimensional data
• Developed inference method to estimate missing data o Established methods to efficiently estimate the mean and covariance of data with missing values o Innovated Gaussian process methods to improve inference 2% over linear interpolation (Python)
• Nonlinear dimensionality reduction
o Identified low dimensional embedding of neural population data via t-distributed stochastic neighbor embedding (Python)
o Applied density clustering to discover target activity via unsupervised methods (Python) Postdoctoral Associate, Massachusetts Institute of Technology 2013 to 2019 Machine Learning: Bayesian inference beyond Gaussian statistics
• Discovered approximate algorithm for cue integration with signal dependent noise statistics o Derived ML and MAP solutions for likelihoods with signal dependent noise (Python) o Engineered efficient updating algorithm to approximate Bayesian performance within 1% (Python)
• Engineered maximum likelihood decoder of point process data streams o Used linear-nonlinear encoding and inhomogeneous Poisson statistics to model encoding (Python) o Executed efficient inference from high dimensional count data based on encoding model (Python) o Deployed nonstationary methods to improve decoding performance 300% over stationary models
• Innovated decoding approaches through analysis of high dimensional data geometry o Leveraged PCA to visualize data and localize regimes of linearity in nonlinear data (Python) o Innovated inference strategy based on locally linear operators on nonlinear embedding (Python) AI: real-time inference via RNN solutions
• Engineered a recurrent neural circuit to implement inference algorithm o Innovated autoregulation architecture to enable learning from sparse inputs (Python) o Validated network performance against a wide variety of data sets (Python) o Developed training strategies to optimize performance according to noise statistics (Python) SETH W. EGGER, PhD Durham, NC 916-***-**** *********@*****.***
/www.linkedin.com/in/seth-egger
/github.com/swegger
Epic Games Seth W. Egger 7.17.25
TECHNICAL SKILLS
• Programming Languages: Python, C, Julia, MATLAB, SQL
• Libraries: NumPy, Matplotlib, Pandas, TensorFlow, SciPy
• High-Performance Computing (HPC) and Parallel Processing: SLURM, Parallel Computing Toolbox
• Version Control Systems: git, GitHub, GitLab
• Operating Systems and Scripting: Bash scripting, Linux, Mac, Jupyter
• Data Types Studied: time series, tabular, behavioral, spectral, imaging, EEG, MRI, MEG, CT DATA SCIENCE AND AI TOOLBOX
• Machine Learning and Data Analysis Techniques: optimization, clustering, kernel density estimators, regression, classification, PCA, SVD, unsupervised learning, neural networks, data wrangling
• Deep learning: RNNs, Convolutional neural networks, LLMs, LSTMs/GRUs, Transformer Networks, Adversarial networks, Prompt engineering, backprop
• Statistical Analysis: hypothesis testing, probability distribution modeling, Bayesian analysis, hidden Markov models, curve fitting, signal detection, bootstrap and permutation, correlation, partial correlation, parameter estimation, cross validation, Gaussian process regression, information theory
• Control Theory: nonlinear dynamics, dynamical systems, linear systems analysis, Laplace transformation, linear-quadradic-Gaussian
LEADERSHIP AND TEAM WORK
• End to end project management
• Lead conception, experimental design, data collection, analyses, interpretation and communication of data narratives for 8 journal publications (6 lead, 2 co-authorships)
• Over 270 combined citations (Google Scholar)
• Diverse roles in project collaboration
• Offered both experimental and analytical expertise, yielding 3 co-authored publications with both junior and senior scientists
• Comprehensive mentorship in experimental design and analysis
• Mentored 5 junior researchers in the successful completion of conference presentations, publications, and thesis projects
• Elevated research capabilities of juniors through technical assistance with Bayesian inference, neural network engineering, software development, and hardware troubleshooting
• Organized journal club discussions and presentations on computational and systems neuroscience
• Designed to integrate expertise of different research groups
• Focused on constructive, critical feedback on written and oral research data presentations
• Organized collaboration for stakeholders across labs, universities, and public/private organizations, resulting in 3 journal publications
INVITED LECTURES
• “Neural structure of a sensory-motor decoder for motor control.” University of Western Ontario, London, ON, Canada. (2022)
• “Computation through the control of neural dynamics.” University of Pittsburgh, Pittsburg, PA.
(2023)
• “Control of dynamic activity across neural representations.” New York University, New York, NY.
(2023)
• “The neural circuit basis of flexible action policies.” RIKEN Center for Brain Science, Wako-shi, Japan. (2025)
EDUCATION
Doctor of Philosophy, Neuroscience, University of California, Davis, CA 2006-2013 Bachelor of Science, Psychology, University of California, Davis, CA 2001-2005