AI + CRM and Recommendation Recall & Ranking Practices Presented at ML‑Summit 2021

From April 16‑17, 2021, the global machine‑learning conference ML‑Summit 2021 was held at the Westin Beijing Hotel, featuring two AI sessions organized by the 58.com Technical Committee AI Sub‑Committee.

Session 1: AI + CRM – Improving Enterprise Performance

The talk, delivered by Jan Kunlin, head of 58.com TEG AI Lab, described how AI technologies are integrated into the company’s CRM system to create a new revenue‑growth engine, including machine‑learning‑based opportunity allocation, intelligent voice quality inspection, and voice‑bot‑driven call efficiency. Real‑world applications from the “Michigan” direct‑sales model, such as ranking, voice quality inspection, and conversational AI, were showcased to illustrate gains in sales efficiency and results.

Jan Kunlin’s background: senior algorithm architect, AI product planner, and former senior engineer at Tencent, with a master’s degree from the University of Chinese Academy of Sciences.

Session 2: Systematic Recommendation Recall Construction and Ranking Optimization

Presented by Luo Jing, head of 58.com TEG Recommendation Technology Department, the session explained how the company built multiple recall channels (basic, vector, tag, innovative) and evolved from manual configuration to data‑driven and automated closed‑loop strategies. A deep heterogeneous interest model was introduced for ranking, incorporating diverse user behavior sequences to achieve more precise interest modeling across scenarios, resulting in improvements in coverage, diversity, and conversion rates.

Luo Jing’s background: senior algorithm architect, former researcher at MSRA/CRL, and experienced engineer at Tencent and Xiaomi, focusing on search, recommendation, advertising, user profiling, and machine‑learning platforms.

Additional conference poster and images are included below.