A talk recap from the Big Data & Analytics Innovation Summit covering three key challenges Lazada's data science team faced while scaling from 4-5 to ~40 people over three years. The challenges include: (1) balancing manual business input vs. automated ML systems, illustrated by product ranking and manual boosting thresholds validated through A/B tests; (2) managing development speed vs. production stability, emphasizing the cost of technical debt and the need for documentation, testing, and code reuse; and (3) aligning short-term business priorities with long-term data science innovation, addressed through timebox-ed skunkworks POC projects with clear deliverables and leadership sponsorship.
Table of contents
How much business input/overriding to allow?How fast is “too fast”?How to set priorities with business?Conclusion12 Impressions