Oleksii
Berliziev, Fractional CTO and Tech Lead for Founders

With a free call where you walk me through the idea. You don't need a spec or a tech plan, just the problem you're solving.

On the call I ask questions, push back where the scope feels off, and give you a rough cost estimate on the spot. Most founders at this stage are carrying assumptions about what to build or in what order that fall apart under one or two questions, and it's cheaper to find that out now than three months in. You leave the call with a clearer picture of the project and a number, whether or not we work together.

Hourly by default, starting at $28/hour. Fixed price only when scope is genuinely locked.

Most engagements run hourly because real software work has unknowns. Fixed price requires a spec detailed enough that we'd spend weeks writing it before any code ships, and most projects aren't worth that overhead. For larger commitments, I offer discounted hour blocks ([40h] / [80h] / [160h], with a usage window we agree on upfront) and monthly retainers. The free call gets you a rough estimate on the spot, so you know the order of magnitude before you commit to anything. If you're optimizing purely on rate, I'm probably not the right fit. I'll tell you that honestly on the discovery call rather than waste your time.

Yes. Send me what you have and I'll tell you what's there.

I use Cursor and Claude Code every day, so I read AI-generated code the same way I read code I wrote myself. I've spent years taking over existing codebases and modernizing them without freezing feature delivery, which is the same shape of problem with a different smell. Bolt and Lovable projects are slightly different because the stack was chosen for you, so step one is figuring out what's actually editable. Either way, the answer to "can you take this over" is yes.

A rescue usually costs less than half of starting over and finishes faster, because the hard problems your MVP already solves don't need to be re-solved.

Most of what a working MVP cost you isn't the visible code. It's the domain decisions, edge cases, integrations, and user flow choices baked into it. Starting over means re-solving every one of those, plus you ship nothing while the rebuild happens. Taking over and modernizing existing codebases is most of what I've done for ten years, on human-written legacy more often than AI-built, and the pattern is the same: keep what works, rebuild what doesn't, ship features while you do it. The cases where rebuild really is the right call exist, and I'll tell you honestly when I see one. Some codebases are genuinely as bad as you fear. Most aren't. The call tells us which.