The gap between senior engineers and prompt-typers isn't who writes better code — it's who can diagnose what breaks, reason about scale, and know when something is actually production-ready.
Sign in with GitHub →Scenario · Intermediate · 45 min
# before.py — what the LLM generated
def get_orders(user_id):
orders = Order.objects.filter(user=user_id)
result = []
for o in orders:
result.append(o.item.name)# ← why is this slow?
return result
Viveka mentor
“How many database queries does this loop execute for 100 orders?”
AI made code generation cheap. A developer can produce a service, test, component, or migration before lunch. That is useful — but it also makes surface-level output easier to fake.
What has not changed is the work behind senior judgment: noticing the edge case, reading the data flow, seeing the scale failure before production, and deciding whether a fix is safe enough to ship.
Viveka exists for that gap: the space between “the code runs” and “the engineer understands why it is correct.”
Every scenario moves through the same disciplined loop: understand the system, solve the defect, debug the behavior, improve the design, then explain the reasoning.
You're dropped into a real codebase with a real bug report. No hints. Read the code, understand the system.
Fix it in a browser IDE. Tests give you signal; they don't tell you why. That's on you.
Your mentor asks questions — never answers. "What does this variable hold after the second iteration?" You think out loud.
Tests pass. Now make it production-ready. The LLM mentor keeps asking until your reasoning is airtight.
Three questions: root cause, scale, production readiness. Your answers are scored on reasoning depth, not keywords.
Tests prove the fix works once. The artifact shows what you understood: root cause, scale behavior, production readiness, and the concepts you covered or missed.
Share it with recruiters, managers, mentors, or your future self — as evidence of engineering judgment, not just completion.
Verified Artifact
Performance
51 → 1
database queries
Reasoning
8.4/10
depth score
Concepts
“The endpoint was making one additional query per order. It passed with small test data, but production volume would make latency grow linearly with every returned order.”
“Closer to a production war room than any learning platform I've seen.”
Senior Backend Engineer · 13 years experience
“Tests whether you think small picture first or big picture first. Interviews never do that.”
Tech Lead · 9 years experience
“This might be how engineering interviews look in the future.”
Software Engineer · 5 years experience
“No other platform encouraged this level of thinking.”
Software Engineer · 1 year experience
“Made me stop, think, then build. In that order.”
Software Engineer · 3 years experience
“The code is just the beginning. The three questions at the end — that's the real test.”
Tech Lead · 9 years experience
Practice with real scenarios. Learn from a mentor that refuses to spoon-feed. Leave with proof that your reasoning holds up.
Sign in with GitHub →