About
Currently, he is the VP of Engineering at Normal Computing, a generative AI startup. Previously, he built the Data Science & ML Engineering groups from 0 to 1 at Ampersand TV. His work at Ampersand included launching one of the largest scale Bayesian modeling systems in use within industry or academia.
Dan specializes in projects at the intersection of data science and software development, though he frequently consults on either individually.
He teaches data science and software engineering online through oreilly.com and in-person. He is an author and contributor to several open source projects and has published articles in peer-reviewed journals. He is the author of Programming with Data: Python and Pandas, a video course published by Addison-Wesley.
Personal Statement
Dan specializes in projects at the intersection of data science and software development, though he frequently consults on either individually. Applying data science for more than ad hoc analysis requires the development of systems to repeatedly gather data, evaluate models, and output results, an end-to-end process in which he has over a decade of experience.
Past Customers
Ampersand TV | November 2019 - January 2023
Sr. Director of Data Science & ML Engineering
Geode Capital Management | 2007 - 2011
Quantitative Analyst Analyst
Kane Capital Management | 2006
Computational Finance Intern
Williams College | 2003 - 2005
Software Developer
Select Clients:
Microsoft
Arvato-Bertelsmann
General Assembly
Sense
Tufts Medical Center
GroupVisual.io
Sustainable Endowments Institute
Quant5
Major National Sports Association
Projects & Accomplishments
Projects:
Flo Technologies - Flo is an IoT company with a hardware and software solution for monitoring home water systems to prevent damaging and wasteful water leaks.
- Designed and implemented their initial statistical models for calibrating user devices
- Built out the production system and REST API for managing the entire workflow related to user device calibration
- Cloud native with Docker containerization and the ability to horizontally scale to meet variable loads on the system.
Languages & Libraries: python, django, celery, numpyPersistence: PostgreSQLMessaging: redisInfrastructure: AWS: ECS, EC2, S3, Elasticache, RDS
Tripod Education Partners - Started by Harvard University professors, Tripod is the leading source of student feedback for K-12 educators.
- Designed and implemented production system for processing hundreds of thousands of student responses
- Developed customizable reports with programmatic Excel generation
- Built validation systems for ensuring integrity of calculation process
Languages & Libraries: python, R, celery, numpy, pandasPersistence: PostgreSQLMessaging: RabbitMQInfrastructure: AWS: EC2, RDS
Sense - Sense is an IoT company with a hardware device for home energy usage monitoring and identification. As part of prototyping their home energy monitor, Sense needed a way to better visualize the data they were collecting and modeling.
- Designed data visualization for millions of data points over variable time horizons
- Implemented REST web backend for serving data to single page application with data visualization dashboard
- Implemented data ingestion process capable of handling hundreds of thousands of records per second
Languages & Libraries: python, ruby, railsPersistence: PostgreSQLInfrastructure: AWS: EC2
Accomplishments:
Talks
- Round Trip Client-Side COPY for High Volume Postgres Inserts, 2Q PGConf 2017, November, 2017.
- Testing Financial Hypotheses with Open Source, Open Finance NYC, January, 2014.
- Open Source Finance with R, Boston Data Mining, December 2013.
- Hands on Machine Learning, Boston Predictive Analytics Machine Learning Workshop, December 2012.
- Predicting Customer Conversion with Random Forests, New England AI Meetup, October 2012.
- Random Forests Lightning Talk, Predictive Analytics World, October 2012.
- Intermediate Regression Topics, Boston Predictive Analytics Meetup, July 2012.
- Programming with Data: Python and Pandas, ODSC, March 2017.
- Introduction to Python and Pandas for Data Analysis - Private Clients - 2016 - Present.
- Introduction to R, Open Data Science Conference, June 2015.
- Introduction to Data Science and Machine Learning – General Assembly Boston – July, 2013
- Genetic Association Studies - Teaching Assistant - Tufts University Medical School - 2013
- Computation for RNA Sequencing - Teaching Assistant - Tufts University Medical School - 2013
Open Source
- bootES: An easy to use interface for calculating bootstrap effect sizes in R.
- portfolio: Classes for analysing and implementing equity portfolios.
- backtest: The backtest package provides facilities for exploring portfolio-based conjectures about financial instruments (stocks, bonds, swaps, options, et cetera).
- portfolioSim: Framework for simulating equity portfolio strategies.
Publications
- Kirby, K. N., & Gerlanc, D. (2017). Finding Bootstrap Confidence Intervals for Effect Sizes With BootES. APS Observer, 30(3).
- Iyengar A, Paulus JK, Gerlanc DJ, Maron JL. Detection and Potential Utility of C-Reactive Protein (CRP) in Saliva of Neonates. Frontiers in Pediatrics, November 2014
- Daniel Gerlanc and Kris Kirby, bootES: An R Package for Bootstrap Confidence Intervals on Effect Sizes. Behavioral Research Methods, March 2013.
- Kyle Campbell, Jeff Enos, Daniel Gerlanc, and David Kane. Backtests. R News, 7(1):36-41, April 2007
Academic Statistical Consulting
- Critchfield AC, Paulus JK, Farez R, Urato AC. “Abnormal Analyte Preeclampsia”: Do the second trimester maternal serum analytes help us to differentiate different types of preeclampsia? [Submitted to Pregnancy in Hypertension]
- Paulus JK, Switkowski KM, Preston IR, Hill NS, Kari E. Roberts KE. Initiation of a Case-Control Study of Pulmonary Arterial Hypertension in Women. Poster presented at American Thoracic Society Annual Meeting, May 2012.
Testimonials
"Dan has worked on several high profile projects with me. I've found him to be a major asset to the team - extremely detail oriented and excellent at documenting his process, analyses and business-centric takeaways. Dan has an ability to understand the nuances and potential issues in a data set, formulate a plan to account for them and derive extremely valuable findings from his careful work. He's very easy to work with, patient in explaining anything and everything to others and an effective communicator that delivers on time and on budget. I hope Dan and I can continue to work together for years to come."
- Ketan Vakil, KMV Digital
"When I met Dan, it didn't take me more than 5 minutes to see that he is a master at his craft. We have worked with Dan for nearly a year on a number of problems, requiring him to apply a wide array of analytical approaches to solve them. He is constantly pushing himself to stay on top of the latest techniques in this rapidly evolving space. While Dan can dive as deep into the data as we require, he stands out from other data consultants I've worked with by being agile and efficient, delivering within budget every time. This type of work can get very technical very fast, and while Dan can speak to our CTO in what sounds like Greek to me, he also has the sensibility to explain the business value of his work in very practical, no nonsense terms. I am truly enthusiastic to endorse him, and can be contacted to answer any questions."
- Alex Loijos, RentJoy
"Dan did a great job in helping us at Healthrageous to reach our goal of generating a set of slick customer reports...Dan was critical in allowing us deliver customer reports because he created a DB layer and a set of scripts to populate it that transformed from our original data schema into a new schema that had cleaned and aggregated our data so that now it directly mapped to the metrics in the customer reports. Dan was able to work through the complexities of our schema, which had evolved over multiple iterations of our product and therefore contained many quirks and inconsistencies that had crept in as our capabilities and offerings had expanded and grown more complex over time. Dan was undaunted by this complexity as he worked with us and iterated through stages of delivery and testing that uncovered issues we didn't anticipate and therefore couldn't work into the initial design.
Dan was easy to work with throughout our dealings, steadily making progress and effectively communicating that progress and any new issues as they arose. I would happily work with Dan again given the opportunity, which is the best endorsement I can give!"
- Jason Sroka, Healthrageous
Education
Williams College, B.A.
Hobbies
- Improv
- Hiking