From a 30‑Year Coding Journey to AI Infrastructure: Wang Yi’s Story and the Open‑Source Projects SQLFlow and ElasticDL

Wang Yi, a researcher at Ant Financial, turned 40 in 2019 and celebrates 30 years of coding that began with a BASIC‑based learning machine at age 10. His early experiments included modifying a washing machine and writing his first game in Apple BASIC and 6502 assembly.

After stints at Tencent, Google, and Baidu, where he contributed to projects such as PaddlePaddle and Deep Speech 2, Wang joined Ant Financial and focused on AI infrastructure. He launched two open‑source projects: SQLFlow, a “translator” that converts analysts’ SQL queries into Python code to train and serve AI models, and ElasticDL, a TensorFlow 2.0‑based distributed deep‑learning system that intelligently schedules resources and reduces idle time.

SQLFlow bridges the gap between analysts who prefer SQL and modeling teams that write Python, enabling rapid experimentation without requiring analysts to learn Python. ElasticDL improves cluster utilization through elastic scheduling, fault tolerance, and smart parameter selection, addressing the massive data and heterogeneous workload demands of the financial sector.

Wang emphasizes a “practice‑first” philosophy: open‑source, code review, and collaborative development are essential for building reliable, scalable AI systems. He believes that practical engineering, supported by solid mathematical models, drives progress more effectively than isolated theoretical work.

The article also includes links to the open‑source repositories (https://sqlflow.org and https://elasticdl.org) and highlights Wang’s view that open‑source transparency and rigorous code review foster better teams and more trustworthy technology.