Vasiliy Kaminskiy - How Research Teams Can Deliver Higher-Quality Insights Faster | PyData

Learn how JetBrains structures their research process into 9 key steps and implements best practices for faster, higher-quality insights delivery through improved workflows.

Key takeaways
  • Research process at JetBrains is structured into 9 key steps, from request ideation to final presentation and feedback

  • Critical elements for successful research delivery:

    • Clear request management and prioritization based on company/product strategy
    • Comprehensive briefing process to define stakeholder expectations and requirements
    • Structured research design with peer review from experienced data scientists
    • Consistent documentation and artifact storage using research notebooks
    • Regular knowledge sharing both within team and company-wide
  • Project management best practices:

    • Split large projects into smaller blocks/iterations
    • Use templates to capture key information
    • Store all research artifacts (SQL, recommendations, visualizations) in one place
    • Implement proper version control and commenting systems
    • Create interactive dashboards for stakeholder engagement
  • Stakeholder management approach:

    • Identify all explicit and implicit stakeholders early
    • Define clear expectations through detailed briefings
    • Present results through live demos or Q&A sessions
    • Gather both automated and manual feedback
    • Address negative results proactively through clear communication
  • Quality assurance measures:

    • Peer review process for code and methodology
    • Regular retrospective sessions to improve processes
    • Validation of research impact through stakeholder feedback
    • Contribution to company-wide knowledge base
    • Focus on reproducibility and documentation
  • Tools and infrastructure:

    • Centralized research platform (DataLore) for consistency
    • Interactive visualization libraries for better presentation
    • Automated survey system for feedback collection
    • Integrated version control and commenting system
    • Unified environment for all users with consistent dependencies