How a Small Algorithm Team Balances Mature, Breakthrough, and Innovation Projects

Team Planning

At the start of the year the author, a PhD in applied mathematics and technical lead of a 7‑person algorithm team at NetEase Yanxuan, divided projects into three categories: mature projects (already delivering stable revenue), breakthrough projects (high risk, not yet launched), and innovation projects (new ideas to be discovered).

Resource Allocation

He assigned experienced members (P4) to lead breakthrough projects while letting mature projects be handled by senior staff with newcomers, minimizing his own management load.

Problem Discovery

He emphasized validating project feasibility, improving communication efficiency by replacing formal weekly meetings with direct point‑to‑point discussions.

Finding Opportunities

He used data analysis and direct business interviews, including a month‑long rotation in the product team, to uncover pain points that led to a cross‑department efficiency project and a new internal rotation mechanism.

Project Execution

For a welfare‑decision algorithm with a one‑month deadline, he reviewed the model, recruited three additional engineers from other teams, and delivered four models, though the final impact was modest.

Side Projects

He experimented with automated video generation and 3D product scanning, producing demo videos that attracted internal interest and led to the creation of a video‑content technical group.

Business Experiments

He attempted a matchmaking service, first offline and then online, but faced low conversion and scalability challenges.

Takeaways

He shares principles such as tackling difficult problems, embracing failure, defining problems before solving them, and aligning technical work with business value.