Experience

• 

  Self-employed

  Jan 2009 - Aug 2018

  Tutor

  I tutored middle school through adult students (and all ages in between) in physics, math (including calculus and college-level statistics), chemistry, and test prep (SAT, ACT, SSAT, ISEE, etc).

• 

  No name (private venue on fairfax)

  Dec 2014 - Dec 2017

  Front of house audio engineer

  Clients include musicians (Stevie Wonder, Lady Gaga, John Mayer, Nas, Panic! At The Disco, Travis Barker, etc.), comedians (Dave Chappelle, Tig Notaro, Jerrod Carmichael, etc.), and companies (Adobe, Activision, Variety Magazine, United Talent Agency, Ketel One, etc.).Responsible for making great events happen by handling a wide variety of technical issues on a tight timeframe while still remaining calm and pleasant (to avoid unnecessary mistakes and to make life more pleasant for our clients, the musicians I'm helping!). I rigorously check for any possible sources of error to prevent hiccups, and deal with the inevitable mistakes and oversights that still come up as quickly and seamlessly as possible, ideally before the musicians or audience even notice.Duties include planning mix sessions and stage plots, guiding the band and management through sound check to make sure they're happy with their sound, mixing, recording our shows in Pro Tools, patching cables, and setting up microphones.Info on the venue: https://www.yelp.com/biz/no-name-club-los-angeles Show less

• 

  Live nation

  Feb 2017 - May 2017

  Machine learning intern

  Working remotely, independently created and implemented a time series algorithm in R to predict future concert set lists of bands based on past set list data. A description of how it works is attached. You can also view the code and the summary of how it works at https://github.com/gregfaletto/setlistpredictor.

• 

  Ziprecruiter

  Jul 2017 - Jan 2018

  Data analytics research intern

  • Using R, developed a model to predict whether a job seeker will apply to a given job listing to infer preferences for listed perks.• Accessed data via SQL. Chose observational study design and model form. Conceived outcome metric, constructed from raw data.• Final deliverable was approximately 3500 lines of R code and a written report.

• 

  University of southern california

  Jan 2019 - May 2023

  Research assistant, graduate assistant lecturer

  • Full-time data science researcher and lecturer Jan. 2021-May 2023; part-time research assistant Jan. 2019-Dec. 2020.• Designed, coded, and tested novel methods for top venues (International Conference on Machine Learning, PNAS). (1) [ICML 2023] PRESTO estimates rare event probabilities, like probability of purchase after viewing an ad (github.com/gregfaletto/presto). (2) Fused extended two-way fixed effects is a panel data causal inference (difference-in-differences with staggered adoptions) ML method (arxiv.org/abs/2312.05985; under review at Journal of Econometrics). (3) Cluster stability selection is a feature selection method for clustered features (github.com/gregfaletto/cssr-project).• Created a novel recommendation system with a startup. Used matrix completion to estimate factors of an approximately low-rank matrix, and harnessed learned factors in a model estimating click probabilities, improving probability estimation by 5.7%.• Taught 100 students $375,000 worth of courses on analytics in Excel & JMP (SAS); making dashboards; communicating results. Show less

• 

  Google

  May 2021 - Aug 2021

  Data scientist intern

  • Designed, programmed simulation studies in Python to quantify flaw in prior method for estimating A/B test treatment effects. Crafted specific solution from problem description. Created a new method for treatment effect estimation & coded in Python.• Coordinated with team, responding to and incorporating broad-strokes objectives, informal feedback, and formal code reviews.• Reduced bias and MSE of treatment effect estimates by over 99% in simulations, while controlling Type I error rate much better.• Submitted 4000 lines of documented, reviewed Python code to Google codebase implementing method and simulation studies. Show less

• 

  Videoamp

  Mar 2023 - Apr 2025

  • Create methodologies using advanced but interpretable causal inference techniques (including double machine learning, instrumental variables, & difference-in-differences) to estimate lift of ad campaigns on observational & experimental data. Derive confidence intervals, conduct power analyses, & write code for methodologies at production scale (tens of millions of observations). • Inspect data quality using tools like exploratory analysis and feature importance metrics to ensure superior model performance.• Collaborate across teams to better understand data and clients’ priorities to direct my efforts.• Created tool to compare viewing metrics under actual ad schedule vs. alternative to demonstrate the value of targeting.• Produce & maintain documentation. Communicate complex data analysis results to stakeholders (presentations, writing).• Proactive identify problems and clearly communicate early to solve. Identify and communicate blockers.• Use tools including Python (scikit-learn, matplotlib, NumPy, pandas), SQL (Snowflake), GitHub. Show less

  • Intermediate Data Scientist

    Jul 2023 - Apr 2025

  • Part-Time Apprentice

    Mar 2023 - Jun 2023
• 

  Google

  Apr 2025 - now

  Data scientist research, payments