TL;DR
I led the AI development behind Daltix’s A3 Tool, transforming a slow, manual assortment planning process into an automated, AI-powered workflow. I built systems for attribute enrichment, LLM integration, and human-in-the-loop validation, replacing weeks of spreadsheet work with near-instant insights. The result? 12x faster, 5x cheaper planning and a multi-year contract to prove it.
Retailers traditionally relied on slow, manual processes to collect, enrich, and analyze product assortment data. This analog approach made strategic planning time-consuming, error-prone, and costly—limiting how often assortment reviews could happen.
As the lead engineer with full ownership of AI at Daltix, I was responsible for building the intelligence layer behind the A3 Tool: an automated, AI-powered assortment analysis product designed to replace weeks of manual work with instant, actionable insights.
I designed and built core systems for attribute enrichment, LLM integration, and AI-based product modeling. To ensure scalability and cost-effectiveness, I engineered a custom human-in-the-loop framework optimized for high-confidence, low-cost validation. My work enabled the tool to compare SKUs across thousands of attributes—like fruit percentage in jams or scampi sizes—and surface blind spots and market gaps in near real-time.
The customer was able to plan product categories 12x faster and 5x cheaper than before. The impact was significant enough that they signed a multi-year contract post-pilot. In hindsight, I’d have leaned even further into automation—human validation adds value, but doesn’t scale like AI does.