About

Work Experience

• Head of Technology, Keyspire

  Mar, 2026 - Present

  Promoted to succeed the departing Head of Technology; sole technical decision-maker.

  • Built the Keyspire AI Assistant from prototype to paid launch in ~3.5 weeks: an LLM coaching agent (OpenAI Agents SDK, TypeScript, PostgreSQL) with RAG over a 337K-word course library, persistent conversation memory, live listing search, and deterministic deal-analysis tools, at under 1¢ per reply.

  • Made model behavior measurable: a ~130-check live-model eval suite runs on every behavior-changing PR. Used it to cut consistently failing knowledge checks from 24 of 92 to 0 by forcing retrieval and tuning a relevance cutoff from measured score distributions; added prompt-injection fencing, domain allow-listing, and per-user rate limits.

  • Built a review-gated agentic development workflow (parallel coding agents in isolated worktrees, automated review, CI tests, live-model evals) that lets one engineer run several production systems; set the architecture across 8 ADRs and review every merge.

  • Built and run a nightly Python pipeline managing the subscription lifecycle for ~1,000 subscriptions (~$40K MRR): ~975 automated updates over 67 production runs, Stripe reconciliation, and a nightly exceptions list for ops, replacing ~20 hours/week of manual work.

  • Audited 110+ CRM workflows end to end and fixed 65+ bugs.

  • Kept customer, billing, and operations running after three roles (two customer service, one operations support) were eliminated, through automation and process redesign; cut annualized software spend 67% ($286.9K → $93.2K CAD).

  • Built the Deal Analyzer, a full-stack investment-analysis app (React, Express, PostgreSQL) with Google OAuth and CRM-gated access, launch-ready for the ~1,000-member community; its calculation engine already powers the Assistant's deal analysis.

• Data Scientist, Keyspire

  Jan, 2025 - Mar, 20261 year 2 months

  Identified the CRM and data quality as the company's real bottleneck and rebuilt them first.

  • Led the HubSpot to GoHighLevel migration and cleansed 190K+ contact records into a single source of truth.

  • Built CRM automations, including a two-way sync with the 900-member Circle community and a 9,000-lead outbound sequence; directed an external agency across 10+ projects.

• Data Scientist (Capstone Project), St. Louis Blues Hockey Club

  May, 2024 - Jul, 20242 months

  Built predictive models for a confidential classification task with an NHL organization.

  • Raised the core success metric (balanced accuracy) more than 3x through feature engineering, regularization, and hyperparameter tuning, validated against real-world outcomes.

  • Engineered 28 features across multi-source data with extreme class imbalance, missing values, and fuzzy name-matching; compared Gradient Boosting, Random Forest, and neural network models.

• Research Data Analyst (part-time), St. Joseph's Hospital

  Sep, 2018 - Dec, 20191 year 3 months

  • Co-developed methodology and statistical analysis for two psychiatric research papers; built a 150+ medication reference dataset in REDCap.

Skills

Education