Technical Product Owner · AI and data products

Syuzana Tevdoradze

Strategy and growth of AI products

I'm a product lead with a technical background. I work on AI and data products: set priorities, assess the economics and the technical constraints, and drive delivery together with the team. I usually start with focus: the product's function, the main pain, the market and the segment. Then the numbers: will the product pay back, the goals (OKR/KPI) and the roadmap. On that basis the founder decides whether to launch, grow or stop.

Book a free callA first consultation, 20–40 minutes online: you describe the task, I say how I can help and whether you need me at all.

What people come with

Before launching an AI product you need to assess demand, technical feasibility and economics to make the investment decision.
You need a clear development plan: what we build first, what comes next, what it costs, and which metrics will show it is working.
An AI feature is being built into the product, say recognition, recommendations or an assistant. Deadlines slip, quality falls short of what is needed, spending on models and labelling grows. You need to decide whether to continue, change the approach or stop.
It is unclear what customers will pay for, and whether the product pays back once infrastructure, labelling, manual checks and support are counted.
There is no product manager on the team. The founder or the tech lead needs help with research, priorities and managing development.

Ways to work

Two project formats and one inside the team. We start with a free first consultation, then fix scope, timeline and price in writing.

Project work

Assessment of a new product or direction

2–4 weeks

I work out who the product is for and which task it solves. I study the segment, the customer's alternatives and the economics. Together with the founder I define the product's function and the boundaries of the first version.

Result. A development strategy: what we build first and in what order, which metrics and budget, what we test before writing code. And a verdict on the direction: launch, wait, or rethink.

What I need from you. Respondents and access to data come from you.

What's inside

Segment and the alternatives customers use today
Interviews with the audience, including those who didn't buy
One-page economics
First-version boundaries and the order of development

Audit of a live AI product

1–2 weeks

I examine model and data quality, real usage and the cost of running the product. I find what is in the way and what can be left alone.

Result. A rework plan: what to fix, what to drop, in what order, with effort and running-cost estimates. And a verdict: continue, change the approach, or stop.

What I need from you. Access to the system, data and metrics; two or three sessions with the team.

What's inside

What customers actually use and pay for, and what is dead weight
What accuracy the models reach on real data and exactly where they fail
What each customer costs to serve: infrastructure, labelling, manual checks, support
What in the architecture blocks development and in what order to change it

Work inside the team

Part-time product lead

4 or 8 days a month

I take the product work on: priorities, research, requirements for the developers, acceptance of the result, coordination with the team.

Result. The product gets an owner: decisions are made, requirements are written, development follows the plan. When the time comes, I help hire a permanent product lead.

What I need from you. Scope of responsibility and goals are agreed before we start; we begin with one paid month.

What's inside

Priorities and roadmap
Research and metrics
Requirements and acceptance
Decision economics, including the cost of AI

How it works

A first consultation, free, 20–40 minutes: you describe the task, I say how I can help.
A letter with scope, timeline and price. At the agreed scope the price doesn't change.
The work, with a short weekly update.
A final session and a written report.

Before we start, we write down what will count as failure.
In the report I mark separately what the data confirms and what remains an assumption.

Experience

A few results from recent years. Companies are not named; details are in the CV.

AI monitoring for greenhouses: of more than ten detection models we kept the four that worked and rebuilt labelling — average production accuracy went from 0.6 to 0.9, manual annotation fell by about 95%.
Same team: a move to shared services and multi-tenant infrastructure cut cloud cost per customer by roughly 10x and turned the unit economics positive.
Government analytics: an NLP system for 200–250 thousand citizen requests a year — classification, entity extraction, routing; processing became about 5x faster, with a person making the final call.
Mobility data: moving from row-based to vector-based storage took a report from about three days to an hour and a half.

Most often this is computer vision and NLP in production, data platforms, legacy systems and the economics of early-stage AI products.

Download CV (PDF)

Pricing

The rate is from €100 an hour, single tasks from two hours. Project scope is estimated in hours up front and fixed in writing.

Format	Scope	Duration
First consultation	20–40 minutes, free	—
Consultation	1 hour, €150	within a week
Consultation with a written review	from 3 hours, from €350	within a week
Single task	from 2 hours	by agreement
Audit of a live AI product	from 20 hours	1–2 weeks
Assessment of a new product or direction	from 30 hours	2–4 weeks
Part-time product lead	4 or 8 days a month	from 1 month
Full time	full days	on request

The final sum, scope and timeline are fixed in a written proposal after the first consultation. At the agreed scope the sum doesn't change. The proposal states whether any taxes are added.

Contact

Send me a couple of lines about the task; I reply within two working days. Or book the free first consultation straight away.

Book the free consultation

Email: syuzana.tevdoradze2@gmail.com