Experience

Building Angie, Elementor’s MCP-based agentic AI assistant. Focus on making AI agents practical in production: understanding user intent, planning work, interacting with users, and turning requests into real actions.

• Built Angie’s MCP Apps infrastructure from scratch and contributed to the open-source SDK.
• Built Ask and Plan modes, giving users control over how Angie works.
• Built an internal Agents Platform for Support teams to create agents that solve customer issues.
• Built an internal LLM Gateway for provider routing, quotas, billing, and observability.
• Maintain the public @elementor/angie-sdk, used by external developers to build tools and embed Angie.

Shifted from building individual integrations to improving the underlying infrastructure.

• Improved the performance and resilience of the integration infrastructure.
• Migrated services from EC2 and ECS to Kubernetes.
• Split ArgoCD applications into smaller, more scalable units.
• Built and maintained CI/CD flows with CircleCI and Tekton.
• Built Grafana dashboards to make system health and failures easier to see.
• Built and operated infrastructure across ECS, ECR, EFS, S3, and Amazon MQ.

Combined hands-on integration work with technical leadership across the integration team.

• Built integrations with complex HR systems for Fortune 500 clients.
• Deployed and managed applications on Kubernetes with KEDA.
• Ran the integration guild, improving technical processes and knowledge sharing.
• Mentored engineers through technical sessions, pairing, and code reviews.

Built client integrations end-to-end, from design and data pipelines to production deployment.

• Owned features across HiredScore’s microservices from design to deployment.
• Built end-to-end integration solutions and ETL pipelines.
• Reduced the development effort required to onboard a client by 60%.

• Built a TextRank-based text-summarization service for SAP’s internal wiki, from design to production.

• Designed and implemented SQL-based dashboards and reports.
• Monitored performance and costs and provided data-driven insights.