Experience and Education – Andrew Fairless, Ph.D.

Work Experience

Senior Data Science Engineer, SimSpace, 2021 – 2023

Trained Q-learning reinforcement learning (RL) agent for cybersecurity penetration and analyst training scenarios
Visualized agent actions for rapid and intuitive evaluation of agent performance
Redesigned agent architecture and persuaded team to adopt multi-agent vision
Created decision-making framework with 4 adaptability types using multinomial-Dirichlet modeling for a low-data domain
Rigorously tested framework production code and often practiced test-driven development
Interviewed cybersecurity experts and mathematized their expertise to inform agent decisions
Presented ~15 seminars on RL, active learning, cluster similarity, Plackett-Luce model, Bayesian modeling, etc.
Conducted ~50 coding interviews for data science and engineering job candidates from intern to manager levels

Domains: hierarchical condition categories, COVID-19, unplanned readmissions, social determinants of health, pharmaceutical adverse events, care management, workforce allocation, medical vocabulary mapping
Methods: regression, binary and multi-label classification, Bayesian hierarchical modeling, SHAP, supervised clustering
Rapidly devised and implemented predictive model to address a sudden, crucial shift in business needs, despite lack of data
Researched, designed, and implemented multi-stage solution for important client revenue problem
Created customized metrics for comparing ROI of 2 different modeling approaches
Consulted cross-functionally to align project outputs with client needs
Documented projects and answered client questions about modeling approaches and AI social bias
Evaluated, mapped, or created data resources to accelerate data science team productivity and inform clients
Initiated data science seminar series
Successfully encouraged 7 other data scientists to present seminars
Presented 80+ seminars on neural networks, embeddings, interpretability, statistics, study design, causal inference, etc.
Created internal Python package of data science utilities
Advised colleagues on clustering evaluation, sampling, data types, virtual environments, modularized coding, etc.
Wrote company blog post Interpretability and the promise of healthcare AI
Interviewed for company podcast

Predicted hospital-acquired infection scores used in determining Medicare payment rates for hospitals
Modeling included linear regression and random forests with imputation for missing data
Deployed web application to deliver predicted scores to hospital administrators

Medical Science Liaison (MSL), Rheumatology, Bristol-Myers Squibb, 2014 – 2016

Territory: southeast Texas, Louisiana
Discussed company research, company pipeline, clinical practice, and basic immunology with physicians and other healthcare providers (HCPs) at one-on-one and group presentations
Coordinated MSL support of Phase II, III, and IV clinical trials for marketed (Orencia / abatacept) and investigational compounds in rheumatoid arthritis, lupus, psoriatic arthritis, and scleroderma
Developed slide deck and used it to train MSL team on clinical trial responsibilities, rules and regulations, and design
Evaluated potential sites for Phase II, III, and IV clinical trials in rheumatology
Co-mentored and trained new MSL on marketed product and effective communication with HCPs
Trained sales team members in basic and clinical immunology
Co-developed training materials for MSL team

Medical Science Liaison (MSL), Neurology, EMD Serono, 2012 – 2014

Territory: Pennsylvania, upstate New York, Delaware
Established ongoing dialogues with Key Opinion Leaders (KOLs) concerning clinical practice, the therapeutic landscape, scientific advances, and company pipeline in multiple sclerosis
Presented information to patients about clinical trials of approved products (Rebif / interferon β1-a)
Developed and updated company materials for presentation to KOLs and other HCPs
Reported medical developments and competitive intelligence learned from KOLs and congresses
Educated healthcare providers about the disease state and scientific issues using approved materials
Disseminated information on investigator-initiated and other grant programs sponsored by the company
Strategically selected and nominated KOLs with appropriate expertise for advisory boards
Evaluated potential sites for Phase III and Phase IV clinical trials in neurology
Identified and trained 4 speakers; organized 13 speaker programs
2013 President’s Award for overall performance

Characterized development of social behaviors in a rodent model relevant to autism and identified brain structures and neurotransmitters that influence these behaviors
Self-taught and implemented analyses in robust statistics, model selection, hierarchical models, intraclass correlation as well as classical hypothesis testing and linear regression
Statistically analyzed and visualized data with custom R scripts
Collaborated with statistician to re-analyze and interpret archival data
Wrote and published 4 peer-reviewed scientific articles
Presented thesis data at 3 university seminars with 30 – 100 attendees and 6 posters at scientific conferences
Simultaneously coordinated up to 3 major projects in different phases (planning, data collection, writing)
R packages included: ggplot2, lattice (graphing, exploratory analysis); lme4 (linear mixed effects models); irr (intraclass correlations); psych (descriptive statistics); functions from Wilcox 2005 (robust statistics)

Capstone project: Distinguishing reviews about conventional & alternative medicine using textual analysis
Developed decision tree to predict whether Yelp reviews described conventional or alternative medicine
Courses included: R Programming, Regression Models, Practical Machine Learning, Developing Data Products
R packages/tools included: plyr, dplyr (data munging); caret (machine learning); qdap, tm (textual analysis); jsonlite (reading JSON data); rmysql (reading SQL data); data.table (fast data processing); shiny (interactive applications); knitr (documentation, reproducible research); RStudio

Linear and logistic regression, neural networks, support vector machines, cluster analysis, principal components analysis

Linear and logistic regression, support vector machines, cluster analysis, principal components, latent Dirichlet allocation
Python packages/tools included: NumPy, SciPy, Pandas, scikit-learn, Jupyter, Anaconda, virtualenv, GraphLab Create