TL;DR — Our loop is: CV review, intro call, 45-min collaborative coding, paid work-sample project, architecture discussion, references. No algorithm puzzles, no whiteboard binary trees. The goal is to see how the candidate would actually work with us, not how they perform under artificial pressure.
Senior engineering hiring is a signal problem: the best candidates are not actively looking, the test loops most companies use bias toward people good at test loops, and the highest-leverage signal — how they would actually work — is the one nobody bothers to measure.
Our loop
- CV + background review — 20 minutes. We look for production responsibility, not credentials.
- Intro call — 45 minutes. Their work history, our work, what they want next. No technical screen yet.
- Collaborative coding session — 45 minutes. Pair on a small realistic problem — usually a debug task or feature extension in a representative codebase. We talk through trade-offs.
- Paid work-sample project — ~6-8 hours over a week. Real-ish problem from a domain we work in. We pay market rate for the time.
- Architecture discussion — 60 minutes. Their work-sample, but at the design level. What would they change at scale? How would they have approached it differently?
- References — 2-3 conversations with people they shipped with.
What we are looking for
- Production judgment — they have shipped to real users and felt the consequences
- Trade-off literacy — they can articulate why they chose A over B, including what they gave up
- Self-direction — they do not need a PM to translate requirements before they can start
- Communication — they can explain their thinking to non-engineers without losing them
- Curiosity — they care about the problem, not just the solution
What we explicitly do not test
- Algorithm complexity recall — Stack Overflow is real
- Memorized framework APIs — same
- Speed under artificial pressure — production is not a 30-minute whiteboard
- Cultural “fit” as a vibe check — we hire for complementary, not similar
What this buys
A loop that filters for the seniority we actually want. Candidates who would not pass the algorithm screen at a FAANG but who can ship and operate production systems. Lower-variance hires. Better retention.
Frequently asked questions
Why pay for the work sample?
Time is the candidate's scarcest resource. A paid project signals respect, attracts better candidates, and forces us to design problems worth doing. The cost is small relative to a wrong hire.
What about whiteboard coding?
We do a 45-minute pair-programming session on a small, realistic problem — never an algorithm puzzle. The goal is to see how the candidate thinks under collaboration, not whether they remember LeetCode patterns.
Working on something similar?
T-Square architects, builds and operates production systems for learning, AI and custom software products. Talk to a senior engineer for a second opinion.