Video summary

REAL Data Analyst + Data Engineer Interview | Actual Questions & My Answers

Main summary

Key takeaways

Business

Interview business focus summary (Data Analyst / Data Engineer + AI agents)

Role context / strategy

  • The company is building a small team and does not expect one person to know everything (e.g., Power BI, Snowflake pipelines, data fixes, stakeholder work, and AI agentic development).
  • Hiring goal: find someone who can cover:
    • Power BI report development
    • Snowflake data work
    • Ability to ramp on agentic/LLM workflows, even if they haven’t built agents yet
  • Interview stance: they optimize for fit and capability to learn, not only exact skills.
    • A candidate checklist is used by the interviewer to evaluate competencies.

Skills & processes discussed by interviewer (what they need)

Power BI

  • Experience building dashboards
  • Ability to create KPI-based visualizations and stakeholder-ready reporting

Snowflake / data engineering

  • Build pipelines using SQL to move/pull data from sources such as AWS/Azure into Snowflake
  • Support ongoing data operations, including fixing issues in Snowflake data

Stakeholder operations

  • Comfortable with requirements gathering
  • Ability to clarify business problem statements and iterate on report definitions

AI / agentic development

  • They want someone who understands agentic development concepts, even if agents aren’t already deployed.
  • Agents are expected to:
    • Understand data
    • “Present a report” that becomes business intelligence
  • Power BI is framed as something that can work on top of agents.

Learning curve / scaling approach

  • Explicit expectation of a learning curve, since AI agents are new to the team.
  • The team will scale up with limited resources.

Candidate’s demonstrated workflow (operations playbook) — as described

End-to-end Power BI report delivery process

  1. Requirements & KPI alignment
    • Meet with stakeholders (mentioned context: Microsoft Teams / Jira)
    • Stakeholders share mockups and KPIs
      • Examples mentioned: revenue, YoY, MoM, profit, and comparisons like quarterly vs previous period
  2. Data connection & mode selection
    • Connect to data from sources like CSV/Excel, AWS/Azure, MySQL/SQL files
    • Sometimes use live connection, sometimes composite mode
  3. Data QA / validation before modeling
    • Check for duplicates, data types, and null/invalid values (N-values referenced)
    • Remove unnecessary columns/tables for optimization
  4. Modeling
    • Build relationships (one-to-one, one-to-many, many-to-many) based on requirements
  5. Measures & calculated logic
    • Prefer measures over calculated columns
    • Rationale: calculated columns consume storage
  6. Visual design & publishing
    • Choose visuals appropriate to each KPI
    • Publish to Power BI Service via workspaces
    • Manage permissions (permission handling mentioned)

Data cleansing tactics mentioned

  • Duplicate handling:
    • Investigate duplicates and interpret cleanup based on client/domain requirements
  • Imputation options referenced:
    • mean / median / mode
  • Use Power Query (Power BI) to replace/transform values

Snowflake / pipeline work (what the candidate actually did)

Pipeline creation via SQL

  • Candidate stated they created pipelines for each table by writing SQL queries, including CTEs and SQL statements.
  • They clarified they were primarily doing pulling/ETL-like pipelines, not creating tables (based on prior work).

Snowflake vs MySQL practical difference (candidate explanation)

  • Keep MySQL for transactional day-to-day operations
  • Use Snowflake as the analytics / warehouse layer so analytics don’t disrupt transactions

Frameworks / playbooks explicitly used

  • Checklist-based hiring (interviewer framework)
    • Interviewer uses a checklist to decide whether a candidate has/doesn’t have required competencies.
    • Decision is delegated to another person/team; interviewer focuses on publishing checklist signals.
  • Data modeling relationships
    • Relationship types explicitly named: 1:1, 1:M, M:M
  • Power BI modeling consideration
    • Preference rationale: measures vs calculated columns (storage/performance implication mentioned)

Key metrics / KPIs (business specifics captured)

  • Revenue
  • Year-over-year (YoY)
  • Month-over-month (MoM) / previous period
  • Profit
  • No specific numeric targets (e.g., revenue growth %, CAC, churn) were stated.

Concrete examples / case-like scenarios mentioned

Power BI project scenario (candidate)

  • Stakeholders provide KPIs and mockups
  • Team gathers requirements
  • Connect sources (e.g., CSV/Excel, AWS + SQL/MySQL)
  • Validate duplicates/data types
  • Model relationships
  • Define measures
  • Build visuals
  • Publish to Power BI with permissions

SQL competency test example

  • “Class of students” problem:
    • Group by birth month
    • Count people per month
    • Use GROUP BY + count, order by birth month

AI agents example (interviewer’s explanation)

  • A small agent built by another team:
    • Input: copy/paste an email
    • Output: a response draft
    • Agent “scope” defined via documentation
    • Agent can access a data bank and may require user review/edit before sending

Topics not covered

  • Sales/marketing: No go-to-market, sales funnel metrics, or marketing strategy content appears.
  • Investing/markets: Not discussed; focus remained on execution and team composition for analytics + AI agents.

Presenters / sources

  • Interviewer: Aditya (referred to as “Adita”)
  • Presenter / Candidate: Madu (Mad Mangla)

Original video