epistemic perspectives

Informing engineering from the cognitive sciences

Larry Muhlstein

Work Experience
Machine Learning Scientist
  • Redesigned and supervised construction of anti-spoofing systems across all of the company’s location-based augmented reality products using a novel deep learning algorithm
  • Designed customer purchase prediction system using LSTM sequence models
  • Various data analysis and prototyping projects
  • Designed distributed reinforcement learning system for internal products
Head of Machine Learning and Data
  • Designed and built production machine learning and algorithmic matching systems for the company’s core business
  • Contributed significantly to the company’s product design and mission
PhD Research Intern
  • Designed and implemented a novel deep neural network based recommender system to replace Snapchat’s current Discover recommendation engine
  • Drafted and filed a three-part provisional patent on my work
  • Presented my work to the research team and to CTO Bobby Murphy
University of California San Diego
PhD student in Cognitive Science
GPA: 3.93/4.00 Advised by Julian McAuley and Virginia De Sa MS degree granted
Case Western Reserve University
8/2010 – 5/2014
B.A. Custom Major in Embodied Intelligence and Major in Cognitive Science
GPA:   3.69/4.00 4.0 GPA in cognitive science courses 3.87 GPA in computer science courses Minors in Artificial Intelligence, Computer Science, and Mathematics
New College, University of Oxford
1/2013 – 7/2013
Visiting Student
GPA:   3.75/4.00 Computer Science and Philosophy
Contextual RNNs for multimodal learning
  • Building novel LSTM-based recurrent neural network to support joint inference over sequential and contextual variables
  • Using Tensorflow to implement model for large-scale multivariate workout dataset
  • Working with Assistant Professor Julian McAuley in the UCSD Department of Computer Science
Learning Theory for Domain Sensitivity and Transfer
Primary investigator
  • Developing computational learning theory to account more cleanly for inference in real-world problems where we have knowledge about the domain
  • This provides a clear basis for optimal transfer learning
Context-Sensitive Memory Augmentation for Writing
Principal Researcher
  • Built web app to suggest content you have produced that is relevant to your current writing task
  • Uses Latent Dirichlet Allocation NLP algorithm to represent documents in terms of topical composition
Epistemic Models of Pragmatics and Communication
Principal Researcher
  • Derived Bayesian model of pragmatic inference to account for degrees of knowledge and belief
  • This model accounts for a number of observed phenomena not predicted by previous models
Recursive Bayesian Models of Context Coordination
9/2014 – present
Primary Researcher
  • Worked closely with professors Roger Levy, Michael C. Frank, and Christopher Potts to develop models of relevance inference in utterance comprehension
  • Designed and implemented experiment to elicit human judgments corresponding to predictions
  • Prepared detailed model comparison analysis to make an informed decision about the path to pursue
  • Presented poster at XPRAG 2015
Computational Models of Neural Spike-Time Reliability
5/2012 – 4/2014
Primary Researcher
  • Derived and implemented stochastic Runge-Kutta model of Hodgkin-Huxley neurons augmented with ion-channel subunit-noise to improve both the validity and the runtime of the simulation
  • Designed novel spike time visualization to avoid parameter choice bias in peristimulus histograms
Other Positions
UCSD Graduate Student Association
Departmental Representative
Representing the UCSD Department of Cognitive Science at UCSD Graduate Student Association Meetings
UCSD Interactive Cognition Lab
Graduate student
Developed a novel computational account of how we infer what people mean by what they say and tested it via experiments and simulations. Designed and tested Bayesian models of how people use communicative feedback to verify that their concept structures are in accord with those of their conversation partner. Supervised undergraduate student in building interactive visualizations of the epistemic relations in human communication.
UCSD Computational Psycholinguistics Lab
Graduate Student Researcher
Worked with professors Roger Levy, Michael Frank, and Christopher Potts to develop a novel computational model of how we coordinate on the context relevant to linguistic interpretation and developed experiments to test these models against human behavior. Presented original research to the UCSD psycholinguistics community. Supervised undergraduate artist in creating stimuli to run psycholinguistics experiments.
UCSD Department of Cognitive Science
Graduate Student Representative
Organized graduate student meetings Attended departmental faculty meetings
Crazy Raccoons Inc.
Director of Data Science
Designed a collaborative filtering recommender system to suggest interesting questions to users of their question answering app.
Tech Corps Ohio
Techie Camps Instructor
Co-taught a number of one week technology camps for elementary and middle school children. They covered web development, simple app development, programming in MIT Scratch, and programming in CMU’s Alice.
Brain, Mind, and Consciousness Lab, Cleveland, OH
5/2011 – 8/2012
Research Assistant
  • Helped scan participants in fMRI
  • Wrote software to process data, create cortical surface visualizations and produce activation graphs
Awards and Honors

Nominated by UCSD for the 2017 Microsoft Research PhD Fellowship

Led a team of four to 3rd place in the IEEE Brain Computer Interface Hackathon

Nvidia academic hardware grant

Glushko Travel Award

Published in Oxford/Cambridge poetry journal

Four times published in Case Reserve Review poetry journal

SOURCE grant for independent research

Case Alumni Association Junior/Senior Scholarship

Bank of America Joe Martin Scholarship

Case Western Trustee’s Scholarship

National Merit Scholar

Data Science in Practive
Spring 2017
Graduate Teaching Assistant
Introduction to Research Methods
Winter 2016 and Fall 2016
Graduate Teaching Assistant
Distributed Cognition
Fall 2015
Graduate Teaching Assistant
Cognitive Consequences of Technology
Spring 2015
Graduate Teaching Assistant
Minds and Brains
Winter 2015
Graduate Teaching Assistant
Introduction to Programming in MATLAB
Fall 2012
Undergraduate Teaching Assistant
Technical Skills

Languages and technologies:

Python         Pytorch Spark
Beam/Dataflow Tensorflow Airflow
GCP SKLearn Keras
MATLAB           Java SQL
Lisp             Unix             R
LaTeX           C     HTML/CSS


Domains and techniques:

Machine learning

Bayesian statistics/modeling

Natural language processing

Data structures and algorithms

Human-computer interaction

Human-centered design

Technical writing