The Problem: Realistic Practice Existed, But Not Affordably
Short answer: I was job hunting, needed to practice system design interviews under realistic conditions, and the AI mock interview platforms that could do that well were priced at $20 to $50 a month, an amount that's trivial for a company expense account and genuinely difficult to justify for someone who is unemployed or between offers. So I built ArchWyse: 36 problems free forever, everything else for $3 a month.
Here's the longer version. A few years into my career, I started prepping for system design interviews at larger companies. I did what most engineers do: read the books, watched the YouTube breakdowns, memorized the vocabulary for sharding, consistent hashing, and CAP theorem. I could recite the theory fluently.
Then I sat down in an actual interview and froze. Not because I didn't know the material, but because I'd never once practiced explaining it out loud, under time pressure, to someone asking follow-up questions I couldn't predict. Reading about system design and performing system design live are two completely different skills, and nothing in my preparation had trained the second one.
So I went looking for somewhere to practice. Not another course. An actual mock interview, on demand, whenever I had an hour free.
What I Actually Needed as a Candidate
The requirements were simple:
- A realistic system design prompt, not a toy question
- Something that could ask follow-up questions the way a real interviewer would, not a static checklist
- A way to actually draw and narrate a design, not just type answers into a text box
- Feedback that told me specifically what was weak, not just a pass/fail score
- The ability to do this at 11pm on a Tuesday, repeatedly, without booking a slot with another person
Platforms that offered this existed. They worked reasonably well. But they were priced for someone already earning a senior engineer's salary, not for someone actively trying to get one. When you're unemployed, freelancing between contracts, a student, or early in your career in a market where a US-denominated monthly subscription is a meaningfully larger share of your income, $25 to $47 a month for practice stops being a rounding error and starts being a real decision.
The irony: the people who most need affordable, repeatable system design practice are disproportionately the people for whom a premium monthly subscription is the least affordable. Interview prep pricing tends to be built around the assumption of a working professional's budget, not a job seeker's.
Why AI Interview Practice Tends to Be Priced High
This isn't a criticism of anyone; running an AI interviewer with voice, whiteboard support, and follow-up reasoning is genuinely expensive. Every practice session involves real-time speech processing and a capable model doing sustained multi-turn reasoning over 30 to 60 minutes. Add hosting, and the per-session cost is not trivial.
Most companies building these tools also have investors, teams, and growth targets to hit, so pricing gets set to cover costs and generate margin. That's a completely reasonable business model. It just wasn't a model I could personally afford to be on the paying side of while I was the one job hunting.
I didn't need a business. I needed a tool. So I built one, sized to what it actually costs to run, with nothing stacked on top.
No credit card, no trial countdown, no "book a demo." Create an account and start a full AI-proctored system design mock interview in under two minutes.
Start practicing free →What I Built Instead
ArchWyse is an AI-proctored system design mock interview platform. You get a real problem, an AI interviewer (ARIA) that asks you clarifying and follow-up questions the way a human interviewer would, a whiteboard to actually sketch your architecture, and structured feedback across the same six dimensions strong human interviewers evaluate: requirements gathering, high-level design, APIs and data model, scalability and reliability, trade-offs, and communication. If you're not sure what that experience actually looks like, we broke it down in what is an AI system design mock interview?
Functionally, it's built to be a real substitute for finding a senior engineer friend willing to spend an hour running you through a mock interview, minus the scheduling, the favor-owing, and the limited number of times you can reasonably ask.
The difference from most of the category isn't the technology. It's the pricing model.
The Pricing, in Full, With No Fine Print
This is the part I want to be completely direct about, because vague "affordable pricing" claims are usually marketing. Here's exactly what ArchWyse costs:
$3 a month is roughly what the AI inference and hosting cost per active user actually runs. It isn't a discounted "launch price" that will quietly climb once I have your card on file. I'd rather keep it close to cost and have people actually use the platform than price it like a premium product and price out the people who need it most.
Compare that to what an hour of live mock interview practice typically costs on interview coaching marketplaces (often $100 to $200 per session), or what a monthly AI interview practice subscription commonly runs elsewhere in this category (frequently $20 to $50 a month). At $3 a month, you can practice daily for a full month of interview prep for less than the cost of one coffee run.
Why I Modeled This on LeetCode, Not on Interview Coaching
LeetCode didn't succeed because coding interview practice was a new idea. Practicing algorithm problems existed long before LeetCode. It succeeded because it removed every point of friction between "I want to practice" and actually practicing: no scheduling, no waiting for a partner, no meaningful cost barrier, just a problem and a place to solve it, immediately.
System design interview prep never got that treatment. It stayed gated behind human mock interviews you had to schedule, or AI platforms priced as premium software. I wanted ArchWyse to work the way LeetCode works: you show up, you get a real problem, you start. If you outgrow the free tier, upgrading costs less than a streaming subscription, not a decision you have to budget for.
That's the whole thesis. Not "cheaper than the competition" as a positioning strategy, just pricing that reflects what it actually costs to keep the lights on, so cost stops being the reason someone doesn't practice. For a practical routine on turning that access into actual interview readiness, see how to practice system design interviews, not just study them.
Frequently Asked Questions
How much does ArchWyse cost?
36 problems are free forever, with no card required. Everything else, 27 additional problems plus every new one added going forward, is $3 a month. You can cancel anytime.
Why is ArchWyse cheaper than other AI system design interview practice platforms?
Because it wasn't built to be a venture-scale business. It was built by someone who needed a practice tool and priced it to cover server and AI costs, not to generate margin on top.
Is there a free version of ArchWyse?
Yes. 36 system design problems spanning Junior, Senior, and Staff difficulty are free forever, with the full AI interviewer experience: voice, whiteboard, and detailed feedback across six evaluation dimensions.
How many system design practice problems does ArchWyse have?
63 total. 36 are free. The remaining 27, along with every problem added afterward, are included in the $3/month Unlimited plan.
Who built ArchWyse and why?
An engineer preparing for system design interviews who couldn't find realistic AI mock interview practice at a price that made sense for a candidate rather than a company. ArchWyse exists to close that gap, the same way LeetCode closed it for coding interviews.
36 real system design problems, free, right now. No trial clock, no upsell email, no card on file. If you outgrow it, the rest is $3 a month.
Start practicing free →