Three ways to work together.
I work with companies that need custom software with AI built into the architecture — not added as a feature later. Applications, internal platforms, critical operational systems. Three formats depending on where your product is: from defining what to build, to designing and building the full system, to adding AI to the platform you already have.
Custom AI-native software
For companies building software with embedded AI — a new application, an operational platform, or a complete system — that need it designed for real-world use from day one, not patched after launch.
I design and build the full system end-to-end: use cases and acceptance criteria, AI interaction patterns, technical architecture (retrieval, context design, guardrails, evaluation framework), and full-stack implementation that ships to production.
This isn't a deck of recommendations. You get a working system, documented architecture decisions, and code your team can maintain and extend.
Works equally well for new applications (designed before writing code), internal platforms, or products growing by adding AI from day one.
Ideal when you have a specific build you want to get right the first time.
Let's design it →AI layer on existing software
For companies with their own platform already running who want to add AI inside the product, not as a separate module or feature bolted on top.
I add the AI layer designed inside your current system: where it fits, how it connects with your existing UX, how it respects your architecture, and how it doesn't break when edge cases show up.
I work in collaboration with your engineering and product team, respecting the decisions you've already made — adding, not rewriting. The AI ends up feeling like a natural part of the product, not a trend-driven addition.
Ideal when your platform is already in production and you want to add AI without breaking what works.
Let's talk →AI Consulting
For companies that need clarity before building, or that already shipped AI and don't understand why users aren't using it. No code yet — I design the path.
Three types of consulting depending on your moment:
- AI Diagnosis: if you don't know where to start, I map where AI has real return in your product or operation, what to build first, and why.
- AI Audit: if you've shipped features nobody's using, I identify what broke trust at each point of the flow and deliver an actionable map of what to fix first.
- Architecture Review: if you're about to build and want to validate the proposed AI architecture before writing code, I review technical decisions, identify risks, propose alternatives.
You get a prioritized findings report, a live walkthrough, and when appropriate, a functional prototype built with the real model — not a static mockup.
It's the lowest-risk way to start working together. Most engagements start here and grow from there.
Ideal as an entry point, or when you've shipped AI that isn't taking off.
Learn more about Consulting →Something more specific?
For specific features, integrations, technical fixes, or work that doesn't fit the three formats above, let's also talk. Custom engagement, defined scope, set in a first call.
Tell me what you need →How engagements typically grow
Most teams start with Consulting. It's the lowest-risk way to see how I work and to identify what's actually broken or what to build first. From there, the path becomes clear:
→ Some teams need a specific build. We move to Custom AI-native software.
→ Some already have a platform and need to add AI. We move to AI layer on existing software.
→ Some need both. We start sequentially, consulting first, then the build.
You don't commit to anything beyond the initial consulting when we start. We define the next step together based on what we find.
What I don't do
Clear boundaries are part of working well together.
I don't train or fine-tune models. That's ML engineering, a different specialty.
I don't sell prompts, libraries, or packaged tools.
I don't do isolated UI design without system context.
I don't replace engineering teams or act as a development vendor.
I don't add AI just because it's trendy. Sometimes the answer is not using AI at all.
Not sure which one fits your case?
Tell me where your product is. The first conversation is free.
Schedule a call →