Same OA format as FDSE I think but onsite was stats focused, hypothesis testing and model evaluation metrics. Behavioral was embedded into every round, that was different. Coding bar felt pretty high.
Software Engineer
Fall 2025
No Offer
MIT · Computer Science
3rd year · 3.7+ · 3 past internships
TechnicalMultiple roundsBehavioralDifficult
The problem decomposition round was unlike anything I'd practiced for. Recruiter call about culture and projects, then an OA with coding, SQL and an API task. Virtual onsite had behavioral woven into every round.
Software Engineer
Fall 2025
No Offer
University of Michigan · Computer Science
3rd year · 2.9-3.2 · 1 past internship
TechnicalMultiple roundsUnexpected Questions
Karat screen with 3 easy problems plus time complexity discussion. the onsite is where it gets interesting, they had a problem decomposition round where you design a chess game and break it into pieces and then a learning interview where they literally drop you into an unfamiliar codebase and you have to navigate it with some guidance. I think my decomp round wasn't structured enough because I jumped into coding too fast instead of planning the architecture first. every round also has 20 min of behavioral questions mixed in so you can't just be a good coder
Forward Deployed Software Engineer
Summer 2026
No Offer
University of Toronto · Computer Science
2nd year · 3.3-3.6 · 0 past internships
TechnicalDifficult
Pymetrics and HireVue then a call with HR about availability, then nothing for 3 weeks got an email saying they moved on and never got a real interview.