AeroLeads people directory · profile

Brian Donhauser Email & Phone Number

Applied Economist, Data Scientist | ex-Amazon at Keystone AI
Location: Seattle, Washington, United States 7 work roles 2 schools
LinkedIn matched
✓ Verified August 2026 3 data sources Profile completeness 100%

Contact Signals

LinkedIn Profile matched
3 free lookups remaining · No credit card
Current company
Role
Applied Economist, Data Scientist | ex-Amazon
Location
Seattle, Washington, United States

Who is Brian Donhauser? Overview

A concise factual answer block for searchers comparing this professional profile.

Quick answer

Brian Donhauser is listed as Applied Economist, Data Scientist | ex-Amazon at Keystone AI, based in Seattle, Washington, United States. AeroLeads shows a matched LinkedIn profile for Brian Donhauser.

Brian Donhauser previously worked as Principal Applied Economist at Keystone Ai and Staff Data Scientist at Flexport. Brian Donhauser holds Doctor Of Philosophy (Ph.D.), Economics (Cert. In Computational Finance) from University Of Washington.

Company email context

Email format at Keystone AI

This section adds company-level context without repeating Brian Donhauser's masked contact details.

Keystone AI

Review company-level records connected to Brian Donhauser before choosing the right outreach path.

Profile bio

About Brian Donhauser

Brian Donhauser is a Applied Economist, Data Scientist | ex-Amazon at Keystone AI. He possess expertise in econometrics, time series analysis, r, risk management, macroeconomics and 16 more skills. Colleagues describe him as "I had Brian as a TA while a student at University of Washington’s Computation Finance program. He took his role very seriously, and was always prepared and available to help. He displayed expertise and depth in whatever topic we were discussing, especially when it came to the more challenging aspects of time series analysis.", "Brian was my course instructor / Teaching Assistant for a course on 'Time Series Analysis' while I was studying at the University of Washington (Comp Fin Program). He did a phenomenal job of teaching the course as well making himself accessible at all times. The course was a lot more manageable thanks to his efforts and feedback. He was exceptionally skilled at econometric and time series modeling and did a great job of helping us understand the subject better.", and "I worked with Brian for over a year at AQR Capital Management. We worked together in the Risk Management group, he as a quantitative risk manager focusing on exposure control (among other things) and I as a software engineer. It was a pleasure working with Brian. He was clear and succinct in his communication regarding program requirements. He was always willing to take the time to explain complex mathematics in a way that made it easy for me to program and to learn more about the business side of managing risk. I have saved some of his hand-written explanations so I can refer to them as necessary. No doubt his time spent as an instructor at the University of Washington came into play in his lucid explications. Brian is also bright and inquisitive, not one who simply accepts things as they appear to be, but one who digs down under the covers to see what is really going on. He enjoys healthy debate on any number of subjects. I really respect and appreciate that when he sees that he is wrong about something, he readily admits it and is happy to have learned something new. He is also not afraid to bring up difficult topics that can lead to thorny issues being resolved, instead of being ignore. He was a real asset to the Risk group and AQR. I would be happy to work with him again if the chance arises."

Listed skills include Econometrics, Time Series Analysis, R, Risk Management, and 17 others.

Current workplace

Brian Donhauser's current company

Company context helps verify the profile and gives searchers a useful next step.

Keystone AI
Keystone Ai
Applied Economist, Data Scientist | ex-Amazon
AeroLeads page
7 roles

Brian Donhauser work experience

A career timeline built from the work history available for this profile.

Principal Applied Economist

Current

San Francisco , Ca, Us

CoreAI• Developed and productionized Hattu: a probabilistic hierarchical demand forecast model of weekly retail vaccine administrations (~11,000 series). Blended reconciliation procedures over an ensemble of simple univariate SARIMAX models. Developed novel solutions for historical stock-outs, short training data, sparsity, irregular seasonality, and business tuning. [R, fable, tsbox]. • Developed and productionized Dagon: a generalized additive model (GAM) for forecasting vaccine administrations over variable future launches. Developed a novel functional form to decompose vaccine interest into long-run, week-of-season, and week-of-year components---solving a confounding problem. Recommended pulling launch date forward to maximize revenue; accepted by client. Also integrated as part of Keystone's core IP. [R, mgcv, splines]

Jan 2024 - Present

Staff Data Scientist

San Francisco, California, Us

Ocean Consolidation Service.• Developed science vision, solution framework, and 3yr roadmap (for 4 science heads) for Gaia: an automated, end-to-end inventory management system. Composed of Hyperion: an inventory recommender built as a dynamic stochastic optimization; Selene: an intermittent consumer demand forecaster; and Elios: a set of logistics component forecasters. Presented and approved at VP level. Stalled at SVP level after dramatic company-wide strategy shift, involving CEO ouster and layoff of 20% of employees, effectively postponing Gaia initiative.• Developed and implemented Container Demand Forecaster: a shipping container demand forecast model covering hundreds of trade lanes at a daily time grain. Handled time-smudged moving holiday impacts and intermittent demand. Would have saved company 0.5 containers per high-volume lane per week. [Snowflake, R, ets, auto.arima, tbats, prophet, nnetar, croston]

Jun 2023 - Oct 2023

Economist Ii

Seattle, Wa, Us

Device Economics, Devices.• Developed and productionized AutoBox: an algorithm automating calibration of discrete choice demand models to market actuals. Built as optimization program with iteratively-expanding box constraints. [Python]• Designed and deployed CalibArch: a standardized JSON format for manually calibrating conjoint demand model estimates to market actuals. Allowed scaling model-informed pricing decisions company-wide. [Python]• Built and delivered Conjoint Demand Models for Cloud Cameras, Streaming Media Players, Tablets. Built as conditional logit and hierarchical Bayesian models. Informed optimal product price. [R, mlogit, mnlogit, RSGHB, PostgreSQL].

Jul 2017 - Jan 2020

Economist I

Seattle, Wa, Us

Topline Forecast, Supply Chain Optimization Tech.• Developed and implemented Prime AutoStabl: an automated method stabilizing forecast revisions. Built as cross-validation procedure over novel and established smoothing methods (multi-pass winsorization, revision EWMA, parameter-locking, adaptive training windows). Defined novel mean absolute revision (MAR) to quantify forecast revision stability. [R, PostgreSQL]• Implemented Prime Mortality: a cohort model of Prime member attrition. Built as modification of Lee-Carter’s SVD mortality model. [R, demography]• Developed and implemented Prime Fertility: A multivariate model stabilizing member driver forecasts. Built as VAR/VECM. [R, tidyverse, vars]• Developed and implemented Prime Flow Error Attribution: a novel method recursively linearizing and additively decomposing top-level forecast errors down to any level of driver hierarchy, back to any historical date. [R, tidyverse]• Co-developed and implemented Prime Flow: a 2nd generation Prime member forecast model. Built as bottom-up, non-linear, state-transition model. [R, forecast]• Developed and implemented Prime λ-SARIMA: the first automated Prime member forecast model. Built as top-down, sequential hybrid, Box-Cox/SARIMA. [R, forecast]

Apr 2015 - Jul 2017

Quantitative Risk Researcher

Greenwich, Connecticut, Us

• Implemented dynamic beta hedging algorithm in equity overlay trade model—reducing in-sample Sharpe ratio by 0.1 across funds. [Excel/VBA, T-SQL]• Re-engineered and automated 75% of overlay trade process—decreasing operational risk, saving 2.5 hours trade execution time daily. [Excel, T-SQL]• Originated and developed generalized algorithm to allocate overlay strategy performance across fund strategies—increasing transparency, decreasing size of code base. [R]• Disassembled, documented, and built dashboard for overlay trade model—teaching 5 firm principals, saving 252 researcher hours annually. [Excel/VBA, T-SQL]• Presented series of 4 one-hour talks regarding theory of equity and volatility overlay hedging algorithms—teaching 18 principals, researchers, and traders.

Jan 2013 - Apr 2014

Course/Section Instructor

Seattle, Wa, Us

• Amath 553, Financial Time Series Forecasting (graduate level)• Amath 540, Intro to Computational Finance and Finmetrics (graduate level)• Econ 582, Econometrics (graduate level)• Econ 200/201/301, Micro and Macroeconomics (sophomore–junior level)

Sep 2003 - Mar 2012

Research Assistant

Seattle, Wa, Us

to Professor Charles R. Nelson, culminating in paper Nelson, C. R. (2008). The Beveridge-Nelson Decomposition in retrospect and prospect. Journal of Econometrics 146 (2), 202–206.

Jan 2006 - Jun 2006
2 education records

Brian Donhauser education

Doctor Of Philosophy (Ph.D.), Economics (Cert. In Computational Finance)

University Of Washington

Master Of Science (M.S.), Mathematics

University Of Washington
FAQ

Frequently asked questions about Brian Donhauser

Quick answers generated from the profile data available on this page.

What company does Brian Donhauser work for?

Brian Donhauser works for Keystone AI.

What is Brian Donhauser's role at Keystone AI?

Brian Donhauser is listed as Applied Economist, Data Scientist | ex-Amazon at Keystone AI.

Where is Brian Donhauser based?

Brian Donhauser is based in Seattle, Washington, United States while working with Keystone AI.

What companies has Brian Donhauser worked for?

Brian Donhauser has worked for Keystone Ai, Flexport, Amazon, Aqr Capital Management, and University Of Washington.

How can I contact Brian Donhauser?

You can use AeroLeads to view verified contact signals for Brian Donhauser at Keystone AI, including work email, phone, and LinkedIn data when available.

What schools did Brian Donhauser attend?

Brian Donhauser holds Doctor Of Philosophy (Ph.D.), Economics (Cert. In Computational Finance) from University Of Washington.

What skills is Brian Donhauser known for?

Brian Donhauser is listed with skills including Econometrics, Time Series Analysis, R, Risk Management, Macroeconomics, Matlab, Financial Econometrics, and Microeconomics.

Find 750M verified contacts

Search by job title, company, industry, location, and seniority. Export verified B2B contact data when you need it.