Online marketplace and ecommerce technology
eBay
eBay operates an online marketplace where consumers, entrepreneurs and businesses buy and sell goods across more than 190 markets. Its offerings span fashion, collectibles and vehicle parts. Magical Listing helps sellers create listings faster, while AI-powered card scanning helps collectors identify cards and understand pricing. Authenticity Guarantee provides authentication for eligible purchases, and Guaranteed Fit helps shoppers find compatible vehicle parts.
Founded by Pierre Omidyar in 1995, eBay is led by CEO Jamie Iannone. Its engineering work includes an internally developed vector similarity engine for matching listings by text and images and generating personalized recommendations from browsing behavior. Described in 2023, the distributed system uses HNSW and ScaNN search backends, with attribute filtering and near-real-time listing updates to support retrieval across a changing inventory.
The company reported 136 million active buyers at June 2026, counting buyer accounts that paid for a transaction during the preceding 12 months. In July 2026, eBay completed its acquisition of Depop from Etsy, adding a consumer-to-consumer fashion marketplace with a Gen Z and Millennial audience. It paid $1.4 billion in cash, including preliminary purchase-price adjustments subject to finalization.
2 talks
Newest firstReviewDebt: a practical framework for scoring every pull request — Sachin Gupta, eBay
AI Engineer World's Fair 2026 · 25:00
1 speaker at AIE
Affiliations reflect their AIE appearances, not necessarily current employment.
Start here
- Agents Need Feature Flags
Learn how canaries, rapid rollback, and kill switches apply to agents, and which operational metrics help track mitigation and audit completeness.
Sachin GuptaAI Engineer World's Fair 2026
- ReviewDebt: a practical framework for scoring every pull request — Sachin Gupta, eBay
Learn how a deterministic review-burden score can inform pull-request escalation thresholds through practical scoring examples.
Sachin GuptaAI Engineer World's Fair 2026
Messages from the stage
Control agent behavior independently
Gupta describes separate controls for prompts, tools, models, memory, autonomy, and sub-agents. Cohort-specific variants and per-turn runtime controls make changes to agent behavior more selective.
Look beyond throughput
ReviewDebt considers diff sprawl, cross-file coupling, weak test evidence, and reviewer fatigue. Gupta uses these concerns to question productivity measures focused only on throughput.
Affiliations reflect each recorded session, not necessarily current employment.

