Washington, DC · nicky.bell@gmail.com · LinkedIn
I’ve spent my career on stubborn problems everybody has learned to live with and nobody is happy about. None of them went unsolved because the math was hard — the messy part is human, which is why my PhD is in social science and not computer science. I live between the product team, the engineers, and the customers, and I hand people the tools to succeed after I've left the room.
Experience
Founder
September 2025 – Present
I founded Powered Analysis to give every high-performing operator the confidence to use AI for data analysis. Through specialized curricula on platforms like Maven, I train business leaders to apply a "Data Analyst Mindset" to AI outputs, focusing on critical thinking over coding and math. Powered Analysis is my laboratory to see what's worth teaching in AI and data — not my exit.
Chief of Staff, Customer Success
November 2025 – July 2026
First Chief of Staff at Keywell.ai as it pivoted from a boutique analytics consultancy to an AI services company for health plans. Reporting directly to the CEO, I served as the voice of the customer as Keywell launched its flagship AI product.
Secured new SOWs with existing accounts that grew overall company revenue by more than 40%.
Oversaw Keywell’s first public webinar and first video education assets for distribution with joint GTM partners.
Deployed internal AI agents, including a resume-screening agent that filtered 3,000+ applications for technical roles and eliminated 50+ hours of manual work.
Analytics Lead
June 2025 – October 2025
First data analytics hire, supporting GTM and customer success activities.
Designed data-driven case studies used by the CEO to pursue a key customer renewal and preserve 11% of ARR.
Managed deployment of new software features that accelerated customer data delivery by 85%.
Founding Director of Data Science
April 2022 – March 2025
Delivered the statistical reporting that directly led to the first FDA approval of an AI algorithm to predict breast cancer risk, including designing the analysis plan and the custom Python/R evaluation libraries.
Improved AI model accuracy by 10% by implementing novel training targets for time-to-event data with interval censoring and systematic missingness.
Owned end-to-end deployment of AWS-native ELT pipelines to ingest and validate 2.1+ million complex medical images and records.
Managed two junior data analysts with a 100% retention rate over 2 years.
Promoted from Associate Director in 2023.
Data Science Instructor
The George Washington University
January 2022 – May 2026
Taught courses in Data Analysis for Journalism and Political Communication. My mission was to prepare future journalists and political professionals to be more responsible consumers and producers of data, especially in the AI era. My teaching materials are publicly available on GitHub.
Graduate Researcher
University of Pennsylvania
August 2015 – December 2022
Designed and ran conjoint survey experiments with 2,000+ participants; conducted more than two dozen qualitative interviews with program participants and stakeholders.
Published recommendations for improvements to Trade Adjustment Assistance, a federal job training program.
Project Manager
November 2013 – July 2015
Ensured successful execution of $300,000+ in grant-funded deliverables for the TRIP Project, coordinating 20+ research partners.
Hired, trained, and managed 30+ part-time research staff; credited co-author on academic publications and in Foreign Policy (online).
Education
Ph.D., Political Science
University of Pennsylvania
2015 – 2022
M.A., European Studies
Universiteit van Amsterdam
2012 – 2013
B.A. (Honors), International Relations
William & Mary
2008 – 2012
Skills
AI & Machine Learning
LLM evaluations · Agentic AI & multi-agent systems · Model evaluation design · Claude suite
Statistical Analysis
Experimental design · Conjoint/survey experiments · Causal inference · Survival analysis · Multilevel/hierarchical models · PCA/factor analysis
Data Engineering
Python · R · SQL · Snowflake · Databricks · AWS · Docker · Terraform · CI/CD · Automated testing
Product & Delivery
AI prototyping · Agile/Scrum · JIRA · Confluence · Notion · Figma