TalkVenue B

Robust & Ease to use: The Design and Evolution of Python-based Primitive Data Types

10/1613:45 - 14:30


Bo Wang

Bo WangAI Engineer in Jina AI

Bo received BSc from Lanzhou University, China and MSc from TU Delft, the Netherlands. Major in multimedia information retrieval.

He used to work for IBM (Amsterdam), Sensara (Rotterdam) as AI Engineer. In 2020, He joined Jina AI (Berlin) as AI Engineer.

He has published 3 long/short papers in ACM Multimedia and related workshops. And represented Jina AI spoke at ECIR (European Conference of Information Retrieval) on Python and Neural search.

His major interest is AI-powered information retrieval (or neural information retrieval) and open source. He is the main contributor of Jina, and one of the Lead developers of MatchZoo.


Every Python developer knows ndarray is the basic data structure of Numpy, DataFrame is the basic data structure of Pandas. Jina, as an open-source project aimed at AI-powered search, we expect users can get familiar with Jina in no time, and become Jina user, even contributors. For this reason, we would like to design an elegant data structure that can help us achieve such a goal. With the release of Jina 2.0, we've received over 10,000 stars on Github, and this can not be achieved without Jina Primitive Data Types. In this speech, I would like to discuss the following topics:

  1. The motivation for designing Jina Primitive Data Types.
  2. Our design principles.
  3. The evolution of Jina Primitive Data Types.
  4. The future directions.

I wish this 30-minute talk could inspire our attendance to design their own data structure with Python.