Joseph Porcheddu

Co-founded an agent-agnostic persistent memory system that enables AI agents to retain and share context across sessions. Built Basic Memory cloud teams to help users and agents collaborate and share AI memory across multiple users and agents, contributing to month-over-month growth since launch. Contributed end-to-end development across core infrastructure including billing, identity/auth, managed Postgres, compute, and object storage. Participated in product, infrastructure, and technical roadmap decisions as a founding engineer.

Owned the company’s Order Management System and the system of record for refunds, exchanges, fulfillment, and returns across ecommerce, retail stores, and internal customer service channels handling 10k+ daily transactions. Built backend services that power iOS apps for inventory browsing, virtual dressing rooms, and checkout support for 90+ retail locations. Designed and maintained internal services used daily by customer service, retail, and warehouse teams to fulfill orders, process refunds/exchanges, and manage inventory. Partnered cross-functionally with design, ecommerce, and retail teams to ship customer-facing features on schedule.

Designed and maintained the microservice infrastructure behind the Pensa Inventory Management System used by retailers and manufacturers to track on-shelf inventory. Built and ran an ETL pipeline moving data from AWS to Microsoft Azure to enable downstream analytics across a hybrid-cloud environment. Implemented a Kafka-based publisher/consumer framework to calculate On-Shelf Availability and Share-of-Shelf KPIs to help detect stockouts and misplaced products early. Built an event-driven change-detection system using Kafka, Python, and Postgres to flag stockouts, misplacements, and new SKUs from near real-time shelf-image data. Mentored new hires and junior engineers.