At Picsart, I build the AI systems behind lifecycle marketing, from multi-agent LLM pipelines and next-best-action models to image-aware personalization campaigns.
I run 10+ A/B and reverse A/B experiments weekly, translate requirements into engineering specifications, and keep releases moving across engineering, design, analytics, and product stakeholders. I also replaced manual reporting with automated Databricks refreshes and built feedback workflows that turn roughly 1,500 monthly submissions into actionable product themes.
I still write the code behind what ships, using Databricks, Python, PostgreSQL, n8n, and AI tooling. My work includes a production multi-agent marketing system that saves 10+ hours per week, a feedback miniapp now widely used in Picsart, and a full-stack restaurant reservation platform with a Gemini-powered RAG concierge.
