About

Building something new. Not ready to talk about the product publicly yet. Short version: AI-native, sits around the kind of cross-functional work that doesn't fit neatly into any one tool.

In active discovery. If you run cross-functional work for a company you care about and have opinions about how this role is going to evolve, I'd like to talk.

Vertical SaaS and fintech platform for Main Street businesses, backed by Redpoint, Atomic, and Haystack. Founded by former Affirm executives. Following extensive market validation across target verticals, the company returned capital to investors in September 2025.

• Drove company-wide transition to AI-native operating model, fundamentally changing how the organization scales, hires, and prioritizes engineering work. Delivered agentic AI system integrated directly into product with intuitive SMS interface. Authored comprehensive technology strategy establishing guiding principles of data as core asset, conversational interfaces, and modular architecture for vertical replication.
• Designed and implemented new card program architecture from ground up, working directly with Stripe and outside counsel to navigate federal and state lending law compliance. Achieved >95% instant approval rates and eliminated operational constraints requiring manual customer onboarding, transforming business from high-touch to self-service and enabling weekly GTM experiment cadence. Over 4 months, executed comprehensive market validation and reached definitive product-market fit conclusions with high confidence.
• Architected complete application stack rebuild in response to developer productivity survey revealing negative engineer satisfaction scores. Established Django/Python/HTMX foundation that enabled new engineers to ship production code within their first week and eliminated technical silos, allowing every engineer to work across the full stack independently.
• Redesigned a broken annual performance review process for a 12-person engineering team by introducing AI-assisted, project-based peer feedback using the SBI method via Google AI Studio. Reduced management overhead from two weeks to two focused days, cut individual feedback time to 15–30 minutes, tripled feedback frequency, and achieved 100% positive team sentiment.
• Authored and rolled out a comprehensive L1–L4 Engineering Career Framework for a 10+ person team, resolving structural ambiguity around roles and expectations, and driving 3 engineer promotions in the first performance cycle following launch.

Joined via the Returnly acquisition. Led data and ML post-merger integration, then took over the Shopping & Discovery organization (50+ engineers across iOS, Android, web, backend, and ML) and site leadership for the Spain office (65+ engineers, analysts, PMs, and designers).

• Extended ML recommender system models to incorporate clickstream event data alongside purchase signals, driving incremental conversion rate lifts of +6.5% and +25.1% across two main merchant recommendation surfaces.
• Built an engineering health scorecard across ~10 Spain teams to surface structural blockers, then executed a targeted reorganization within one month — improving morale, establishing clear domain ownership, and transferring full ownership of Affirm's Search and ML personalization infrastructure to Spain.
• Led end-to-end adoption of GrowthBook as a unified A/B testing and feature flagging framework across Returnly's full tech stack. Restructured a stalled initiative to drive cross-functional adoption, delivering the first two production experiments with +30% and +13% incrementality in shopper engagement.
• Identified a critical data accessibility blocker during post-merger integration and led an end-to-end initiative to deliver multiple domain-specific analytics data marts via an external vendor, on time and within budget — enabling Affirm stakeholders to independently access Returnly data for compliance, financial reporting, and product and growth analytics.
• Identified a critical data reliability risk from unmanaged cross-team pipeline dependencies, secured CTO buy-in, and led a build-vs-buy evaluation resulting in org-wide adoption of Prefect. Guided Engineering and Data teams through full migration of all batch workflows, eliminating silent data failures and establishing a reliable cross-team orchestration standard.
• Identified structural root cause of chronic ML underadoption and recurring executive escalations at Returnly. Designed and championed a Data Science PM role, navigated org politics to place it within the Product org, and led the hiring process — resulting in ML becoming embedded in product planning and a significant reduction in escalations.

Joined as IC reporting to the CTO with a mandate to build the data function from scratch. Grew the organization to 40+ people across risk, personalization, experimentation, and analytics within a ~100-person Series B startup. Acquired by Affirm in April 2021 for $300M.

• Built the company's first ML underwriting model. Became the technical foundation of the core product.
• Conceived and shipped a multi-tenant collaborative filtering product recommender system deployed across hundreds of e-commerce merchants, driving 20–30% aggregate CTR uplift and up to 73% for individual merchants. Navigated organizational complexity through cross-functional influence, and engineered a meta-modelling pipeline to automate model training, evaluation, and deployment at scale — all validated through statistically significant A/B testing.
• Led data due diligence for Returnly's acquisition by Affirm as one of a small, confidential team — personally delivering all required analytics and reports on deadline while continuing to run the 40-person data organization. Built a fuzzy logic OFAC sanctions screening service across 150M+ shopper-address combinations, identified and corrected a systematic false-positive bias in Hispanic names, and delivered documented compliance proof that removed a key regulatory blocker to the acquisition.
• Designed and implemented Returnly's modern analytics stack (Fivetran, Snowflake, dbt, Looker) from scratch, replacing an error-prone Postgres replica with a scalable self-service platform. Enabled company-wide data access across all functions for the first time, while maintaining data integrity through a curated metric catalog and analyst-owned trusted dashboards.

Data Science Manager in the Vistaprint business unit of Cimpress (mass customization world leader). Globally responsible for product recommendation models; ran data mining and text-based ML for customer segmentation.

• Built end-to-end instrumentation for a globally deployed product recommender system processing millions of model servings per day, introducing granular model traces, impression and click tracking, and session-level attribution via Databricks and Parquet on S3. Uncovered ~20% more indirect revenue contribution than previously attributed and unlocked faster experimentation across the personalisation platform.
• Identified a high-propensity customer segment for Vistaprint's new Signage product category by mining business card text data to infer physical storefront ownership — a novel hypothesis validated through NLP, cluster analysis, and proxy metrics. A split-test campaign against traditional RFM segmentation delivered approximately 80% uplift, establishing text mining as a core analytical capability within the organization.

The Boston Consulting Group is a leading management consulting firm. As a Senior Analyst, I worked alongside consultants on cases and performed advanced analyses and built predictive models. Furthermore, I developed analytical tools and prototypes of management information systems. Moreover I coached and trained junior colleagues.

PricewaterhouseCoopers is a leading professional services firm in the areas of assurance, tax and advisory. As a Consultant I performed IT and process audits and subsequently advised my clients how to improve their processes and IT systems. Furthermore, I performed forensic data analyses in order to identify fraud.

Development of mostly web-based applications for clients like ESCP University, LexisNexis Germany etc.