finished the Humana final round last month for a Senior Data Engineer role. 'onsite' is basically virtual now but they still call it that. sharing the breakdown because it's a slightly different shape than a pure SWE loop.
structure
4 x 45-min sessions over one full day on Teams. back-to-back with short breaks. they do it all in one day.
session 1: technical coding
got a SQL problem first. realistic healthcare data schema: members, claims, diagnoses. question was about identifying members with specific chronic condition flags who hadn't had a preventive screening in 12 months. window functions, conditional aggregation. not exotic SQL but you need to be comfortable writing it live. then a short Python problem on parsing nested JSON from an insurance API response.
session 2: system / architecture
how would you design a data pipeline to ingest claims data from multiple payer sources, normalize it, and make it available for analytics and ML features? classic ELT design. i talked about Kafka for streaming, dbt for transformation, Snowflake or BigQuery as a warehouse target, and data quality checks at ingestion. healthcare context: data contracts and upstream SLA variability matter, payer files don't always arrive on schedule.
session 3: behavioral
standard format but healthcare-flavored. questions about working with compliance teams, handling PII/PHI in pipelines, and a team conflict resolution example.
session 4: leadership / culture conversation
met with someone from the hiring manager's chain. less structured. more of a 'here's our roadmap, how would you fit in, what questions do you have' format. they definitely used this to assess maturity and how you talk about ambiguity.
overall
fair process. the day is long but they moved efficiently. got my hiring decision 8 business days after the final round. offer came with a background check contingency.
one note: they use behavioral anchors pretty rigorously. the interviewers were taking notes during my answers. structure your STAR stories cleanly.