Building InterviewMochi: A Solo Founder's Journey Back to Code

Three years in investing. One month building InterviewMochi. Here's why I returned to tech and what makes this more than just an LLM wrapper.

The Three-Year Detour

Three years ago, I stepped away from my software career and dove into investing. It was an incredible journey—market-beating returns, multiple wins, the thrill of analyzing companies and making bets that paid off. By most measures, it was successful.

But something was missing.

I couldn't quite put my finger on it at first. The intellectual challenge was there. The financial rewards were there. But the act of building—of taking an idea from nothing to something people could use—that was gone. And I missed it more than I realized.

The Turning Point

Everything changed toward the end of last year. What started as a simple question—“How big should my Amazon position be?”—turned into something much bigger when I discovered the power of agentic development with Kiro.

Suddenly, building wasn't just possible again. It was easier than it had ever been. The barriers that used to slow me down—boilerplate code, repetitive tasks, context switching—were gone. I could think at a high level and dive into details without losing momentum.

I spent the first two months building marketdly.com, a tool for analyzing market data. It worked well for what I needed, but the regulatory complexity in that space made me realize I didn't want to spend the next year navigating compliance issues.

I needed a different idea. Something simpler to launch, but still meaningful.

Why Voice? Why Interviews?

I started exploring ideas around voice. The original thought was simple: analyze the content, structure, and delivery of conversations. Even in an AI-first era, the human element of voice and communication isn't going away. Voice is still the first medium for inspiring, proving, and leading thoughts.

From there, it was a natural progression to think about where communication matters most. And one area kept coming back to me: interviews.

I've been on both sides of the table for hundreds of interviews over a quarter century. I've seen what works and what doesn't. I've watched talented people stumble because they couldn't articulate their experience clearly. I've seen others land offers not because they were the most qualified, but because they communicated with confidence.

That gap—between what people know and how they communicate it—felt like something I could help with.

Building Alone, Building Fast

This is the first product where I've been the complete decision maker. Technical design, UI/UX, product strategy—all of it. No committees, no compromises, no waiting for approvals.

Going solo is a conscious choice. I'm fortunate to be in a position where I can fund the initial development myself, which gives me the freedom to build without external pressure. No investors to report to, no board meetings, no dilution. Just me, the product, and the users.

This approach keeps things simple, nimble, and fast-paced. I can make decisions in minutes, not weeks. I can ship features the same day I think of them. I can pivot without asking permission.

My strength has always been seeing things at the high level and diving into the details when needed. Debugging across technical boundaries, connecting the dots between frontend, backend, and infrastructure—that's where I thrive. And with agentic development, I could move faster than I ever could before.

In one month of focused work on InterviewMochi, I went from idea to a working product:

• •Voice-first interview practice — Record your answers and hear yourself back, just like a real phone screen. No video, no setup, just practice.
• •Detailed AI analysis — Not just generic feedback. We analyze your pace (words per minute), clarity (filler word frequency), structure (did you answer the question?), and content quality (specific examples, metrics, impact).
• •Personalized questions — Questions tailored to your specific role, experience level, target companies, and weak areas. Not a generic question bank.
• •Practice history — One central place to see all your attempts, track improvement over time, and identify patterns in your answers.
• •Progress tracking — See how your scores improve session by session. Watch your filler words decrease, your pace stabilize, your structure tighten.

Why Not Just Use ChatGPT?

Fair question. Here's what makes InterviewMochi different from just typing into an LLM:

• 1.Voice analysis matters. ChatGPT can't hear you. It can't tell you that you said “um” 12 times, that you spoke too fast, or that your voice trailed off at the end. We can.
• 2.Context persistence. Your profile, your history, your weak areas—it's all remembered. You don't start from scratch every time.
• 3.Purpose-built for interviews. Every feature is designed for one thing: helping you get better at interviews. No distractions, no generic advice.
• 4.Structured practice. We guide you through a practice session, not just answer random questions. You get a flow, a rhythm, a real interview experience.
• 5.Future potential. We're building toward custom models trained on successful interview patterns, company-specific question banks, and adaptive difficulty. That's not possible with a general-purpose LLM.

The tech stack is straightforward: Next.js for the frontend, AWS Lambda for the backend, DynamoDB for data, AWS Transcribe for speech-to-text, and AWS Bedrock (Claude) for AI analysis. Nothing fancy, just tools that work.

Entering Unfamiliar Territory

Building the product was the easy part. I know how to code. I know how to architect systems. I know how to ship.

What I'm learning now is everything that comes after: finding product-market fit, identifying the right user base, figuring out pricing, and developing go-to-market strategies. This is unfamiliar terrain for me.

I think the product solves a real pain point. I think it's useful. But I need to validate that with real users. That's where you come in.

What I'm Looking For

I'm not asking you to buy anything (though if you find it valuable, I'd appreciate it). What I need most right now is feedback:

• •Does this solve a problem you have?
• •What features are missing?
• •What's confusing or frustrating?
• •Would you recommend this to someone preparing for interviews?

Try it out. Break it. Tell me what's wrong. Tell me what's right. Either way, I'll learn something.

Another Fork in the Road

Looking back, the past few months feel like another fork in my career journey. Not back to where I was, but forward to something new: building AI-native products as a solo founder.

I don't know where this leads. Maybe InterviewMochi becomes something big. Maybe it stays small but useful. Maybe I learn what doesn't work and build something better next time.

What I do know is this: I'm building again. And that feels right.

Keep building.