just finished an AIG senior software engineer interview process and the system design round was the most interesting part. posting because the AIG system design interview is pretty different from what I expected going in.
first, context: AIG is an insurance company, not a consumer tech company. their scale is not Twitter-scale. their systems deal with policy administration, claims processing, risk modeling, document management, and integration with legacy mainframe infrastructure. if you walk in trying to design TikTok's feed, you're going to look misaligned.
what they actually asked me: design a document processing pipeline for insurance claims. millions of claims per year, PDFs and scanned images, need to extract data and route to the right teams.
what they were looking for: did I ask clarifying questions about volume, SLA, failure modes how did I think about reliability vs cost (insurance ops teams are cost-conscious) did I mention integration with existing systems or just assume greenfield data validation: insurance docs have strict regulatory requirements around what gets stored, where, for how long
what caught me off guard: they asked specifically about GDPR-equivalent data handling for US customer PII. not a gotcha, but something fintech/insurtech companies genuinely care about that pure SaaS engineers sometimes skip in system design.
time format: 45-55 minutes. they give you the prompt, expect 5-ish minutes of questions from you, then drive the design together. it's collaborative not interrogation.
honest take: if you've only interviewed at product-led startups or big-tech consumer companies, spend time thinking about enterprise and regulated-industry constraints before this one. the problems are more interesting than most of the toy scale problems you get elsewhere.