I turn messy information into decisions: econometric research on 28,000+ NYC buildings, an NBA forecasting pipeline, financial models, and a clickable AI workflow prototype for email.
Selected projects in econometrics, financial modeling, sports analytics, and applied AI. Each shows how I frame a question, work with data, and communicate a decision.
Skidmore College · May 2026
Built a 28,132-building panel from two NYC datasets, cleaning 91,000+ records and deriving expected penalties from statutory emissions caps. Modeled compliance in Stata using fixed-effects linear probability models; found higher-value buildings absorb penalties rather than retrofit, contradicting th
Skidmore College · Dec 2025
Constructed a risk-adjusted 3-year P&L and JV structure (5% royalties, 50% profit sharing) enabling zero-CAPEX entry, forecasting $2.1M maximum net cash flow across 16 locations.
I turn messy information into decisions: econometric research on 28,000+ NYC buildings, an NBA forecasting pipeline, financial models, and a clickable AI workflow prototype for email.
I graduated from Skidmore College in May 2026 with a BA in Economics and a Management and Business minor. My thesis studied NYC building compliance with Stata models; I also built a Python NBA forecasting pipeline and led a South Africa market-entry P&L. For RelayIQ, I designed a clickable AI CRM prototype and documented an email classification workflow. I am seeking entry-level data, financial, pricing, and applied AI roles in NYC.
Experience
Digital Operations Assistant
Stephen Barker · Summer 2024
Education
Skidmore College, Saratoga Springs, NY
Bachelor of Arts, Major: Economics, Minor: Management and Business · Class of 2026
Syracuse Study Abroad, Florence, Italy
Columbia Grammar and Preparatory School, New York, NY
Class of 2022
Skills
Skidmore College · May 2026
Built a 28,132-building panel from two NYC datasets, cleaning 91,000+ records and deriving expected penalties from statutory emissions caps. Modeled compliance in Stata using fixed-effects linear probability models; found higher-value buildings absorb penalties rather than retrofit, contradicting th
Built a 28,132-building panel from two NYC datasets, cleaning 91,000+ records and deriving expected penalties from statutory emissions caps. Modeled compliance in Stata using fixed-effects linear probability models; found higher-value buildings absorb penalties rather than retrofit, contradicting the initial hypothesis.
Merged two NYC datasets, cleaned more than 91,000 records, derived expected penalties from statutory emissions caps, and estimated fixed-effects linear probability models in Stata.
The results challenged my initial hypothesis: higher-value buildings appeared more likely to absorb penalties than retrofit. I documented the model and its limitations in my senior thesis.
Skidmore College · Dec 2025
Constructed a risk-adjusted 3-year P&L and JV structure (5% royalties, 50% profit sharing) enabling zero-CAPEX entry, forecasting $2.1M maximum net cash flow across 16 locations.
Constructed a risk-adjusted 3-year P&L and JV structure (5% royalties, 50% profit sharing) enabling zero-CAPEX entry, forecasting $2.1M maximum net cash flow across 16 locations.
Built the three-year P&L, stress-tested revenue and cost assumptions, and modeled a joint venture structure with 5% royalties and 50% profit sharing.
The proposed structure projected up to $2.1M in net cash flow across 16 locations under the modeled scenario. This was an academic market-entry recommendation, not a launched expansion.
Independent project · Mar 2026
Fit a logistic win-probability model in Python on 5 NBA seasons, validated on a held-out season (Brier 0.204, 70% accuracy vs. 55% baseline); built a live pipeline converting market prices into implied probabilities to benchmark forecasts.
Fit a logistic win-probability model in Python on 5 NBA seasons, validated on a held-out season (Brier 0.204, 70% accuracy vs. 55% baseline); built a live pipeline converting market prices into implied probabilities to benchmark forecasts.
Built the Python modeling and live data pipeline, converted market prices to implied probabilities, and evaluated forecasts against a held-out NBA season.
The held-out model scored 0.204 Brier and 70% accuracy versus a 55% baseline. The project demonstrates prediction, validation, and clear performance benchmarking.
Independent project · May 2026
Designed an AI CRM workflow that turns inbound email into structured, prioritized records for human review. Created the product specification, operator playbook, workflow diagram, and clickable prototype.
Designed an AI CRM workflow that turns inbound email into structured, prioritized records for human review. Created the product specification, operator playbook, workflow diagram, and clickable prototype.
Mapped the intake and review steps, specified LLM classification and scoring logic, and built a clickable interface to make the proposed workflow concrete.
Delivered a reviewable product concept and prototype. Gmail integration and automated classification are part of the specification, not a deployed production system.
Skidmore College · Dec 2022
Proposed a technology and furniture rental program tied to Target Circle to close differentiation gaps, supported by a 5-year Excel financial model with pricing and efficiency assumptions. Presented and defended projections directly to Target executives, earning recognition for strategic clarity.
Proposed a technology and furniture rental program tied to Target Circle to close differentiation gaps, supported by a 5-year Excel financial model with pricing and efficiency assumptions. Presented and defended projections directly to Target executives, earning recognition for strategic clarity.
Built a five-year Excel model for a proposed technology and furniture rental program, including pricing and operating-efficiency assumptions.
Presented and defended the academic recommendation to Target executives. The proposal remained a strategy case rather than a launched Target initiative.
Independent project
Built and published an online portfolio with full post-production workflow (RAW, Lightroom, Photoshop); grew site visits 162% in 90 days via Analytics-driven SEO.
Built and published an online portfolio with full post-production workflow (RAW, Lightroom, Photoshop); grew site visits 162% in 90 days via Analytics-driven SEO.
Built and published tomleslie.photos, managed RAW editing through Lightroom and Photoshop, and used site analytics to refine search visibility.
Site visits increased 162% over 90 days following analytics-informed SEO changes. The photography portfolio remains an ongoing personal project.