Resume — Kaede Johnson

Professional Data Science & Economic Research

Feb 2020 – Jul 2022

New York, USA

Macmillan Publishers · Top 5 global book publisher

Led title selection for a $300K-net-revenue digital pricing program that created dozens of New York Times bestsellers.
Presented original competitor-pricing analysis to the CEO, motivating profitable new discounting behavior.
Halved Amazon advertising cost per dollar earned using Selenium web scrapes.
Co-built an SEO tool and presented its 3–6% pilot lift in Amazon views to the CFO.
Co-designed a $200K-net-revenue pay-per-circulation experiment that led to further PPC distribution.
Briefed non-analytic staff on webpage engagement using Google Analytics.
Automated legacy balance-sheet reports and database maintenance with SQL, JDBC, and Docker.

Jun 2017 – Jul 2019

Washington, DC, USA

Federal Reserve Board · Central bank of the United States

Forecasted GDP, CPI, and interest rates for Australia and New Zealand while reducing mean absolute percentage errors.
Used ex-ante forecasts in R and Stata to justify overhauls of unemployment time-series regression models.
Authored 30+ economic notes for international policymaker meetings; selected samples are available in an economics writing portfolio .
Prepared a presentation for the Board of Governors and developed the material into a published Federal Reserve Note .
Maintained the team website in PHP and forecast pipelines in VBA.

Academic HSG PhD, EPFL MSc, UMD BSc

Feb 2026 – Present

St. Gallen, Switzerland

Researching how people understand and interact with large language models.

Sep 2022 – May 2025

Lausanne, Switzerland

Personalizing AI Text Simplification: MSc thesis using SFT, DPO, Hugging Face TRL, and JavaScript. Publication · GitHub
Customized AI Tutor: SFT, DPO, RAG, PyTorch, and Hugging Face. Project paper
LLM Semantic Search: prompt engineering; ranked in the top 10 of 122 projects. Project · course showcase
Goodreads Recommender System: neural graph collaborative filtering with PyTorch. GitHub
SBB Trip Planner: Connection Scan Algorithm and PySpark. Demo · GitHub
Visualizing European Political and Social Views: interactive JavaScript data visualization. Project · GitHub
NLP for Memoirs and Authorship Classification: LSA, LDA, FastText, NLTK, SVM, and spaCy. Memoirs repository · authorship repository

Sep 2013 – May 2017

College Park, MD, USA

Soft skills Leadership & Languages

Executive communication: Presented an SEO tool to executive leadership at the Holtzbrinck AI conference and regularly translated analytical work for non-technical audiences.
Mentorship: Co-supervised a Macmillan intern’s relational database project, providing project guidance and technical support.
Teaching: Taught machine learning to 30 HSG students, macroeconomic theory to five incoming Federal Reserve researchers, economics to 20 high school students, and English to 20 adult learners.
Languages: Native English and B1 German.