datalabz

Applied engineering for transactional and data-heavy businesses.

We ship our own software and take a few client engagements per year on AI-augmented systems where correctness and operability matter more than novelty.

  • Distributed delivery since 2018 Seven years building real systems under confidentiality.
  • Shipping our own products The latest is a thin SQL gateway that lets LLMs answer questions on operational data without seeing it.
  • Selective by design 4–6 clients per year, engagements between $50k and $500k.
Get in touch

How we work

We work with 4–6 clients per year. Most engagements are 6–12 weeks of focused work to ship one thing well — a real-time data layer, an integration backbone, an AI-augmented internal tool. Some grow into longer relationships; most don’t, and that’s fine.

Inside our delivery, AI is a tool we use heavily — not because it’s trendy, but because it actually compounds engineering output for a team our size. Inside what we ship, AI shows up only where it earns its place: decision support, structured generation, retrieval over private data, agentic workflows where deterministic alternatives don’t exist.

We don’t pitch. If the work isn’t a match, we say so and recommend someone else. If it is, we send a one-page proposal within a week.

Recent work

  • Real-time integration layer between a legacy gaming platform and a modern CRM stack — ~2M events/day, sub-second latency.
  • Operational data pipeline for a regional payment processor — settlement reconciliation in minutes instead of overnight batch.
  • AI-augmented analyst toolkit for an iGaming operator — natural-language reports over a curated view layer, audit-logged end-to-end.
  • Anomaly detection on transactional flows for a fintech operator — saved a seven-figure annual fraud exposure.

Placeholder examples — replace with real, anonymized work before launch.

Section 2

What we ship

We build and run our own software. Each one came from real client work and stayed because we wanted to keep building it.

Lead product

AI report gateway (working name — confirm marketing name)

Thin SQL gateway that lets LLMs answer questions on operational data — without giving the LLM access to the rows.

For
Transactional businesses (iGaming, finance, payments) where internal analytics is gated by SQL fluency and audit requirements.
Detail
Hierarchy isolation via session-scoped Postgres variables; the LLM sees the schema and a curated view layer, never the row data. Every generated SQL is audit-logged before execution.
Status
Shipping — in production with a design partner.
Read the spec →

Single product card by design — better than padding with vaporware. Add a second card only when a second product is real.

Section 3

Thinking

A small number of essays we cared enough to publish. Each takes a position. Each stays up indefinitely.

  1. 01

    Your internal data is worth more than the AI you put on top of it

    Most companies have terrible internal data and great external AI. The leverage is in the data layer — the AI is the easy part.

    May 25, 2026

    →
  2. 02

    What 'AI-native' actually means once you ship to production

    If you can't describe how you'd debug a bad model response at 3am, you don't have a production system.

    May 18, 2026

    →
  3. 03

    On-prem AI in regulated industries: the actual constraints

    'We'll just run an open-source model on-prem' is the wrong starting point. The real constraints aren't OSS vs cloud.

    May 10, 2026

    →

Section 4

Careers

We hire slowly and rarely. The roles below are open because we have specific work for someone to own — not as a generic “we’re always hiring” pipeline. If you see your name on one of these, write to us directly.

Distributed (Yerevan headquarters, team across timezones). We expect senior judgment, not 9-to-5 attendance. Smallest commitment is 6 months full-time on a single engagement. No staff augmentation, no agency contracting, no “junior + lead” structures.

01

Principal AI Systems Engineer

Own the end-to-end design of AI-augmented systems we ship to clients and internally. Decide whether a problem needs a model at all, what evaluation harness goes around it, where it sits in the data layer, and how it gets debugged in production.

You’ve already done

  • Built and shipped a production AI feature that survived contact with real users.
  • Designed an evaluation pipeline that caught a regression before it landed.
  • Worked across the stack from data layer to user-facing surface.

You’re not the right fit if

You've only worked with managed AI APIs, you've never been on call for what you built, or you describe yourself as a “prompt engineer.”

Compensation

Equity + competitive base. Specific numbers in conversation.

02

Lead Engineer — Transactional Platforms

Lead architecture for real-time transactional client engagements. Payments, settlement, exchanges, gaming — anywhere correctness, latency, and operability all matter at the same time. Own the technical relationship with one or two design-partner clients at a time.

You’ve already done

  • Designed a system that processed >1M transactions per day in production.
  • Owned an incident that involved money or a regulator.
  • Made a hard CAP / consistency-vs-availability call and lived with the consequences.

You’re not the right fit if

Your experience is mostly CRUD apps, B2B SaaS, or e-commerce. We respect that work; it's not our wedge.

Compensation

Equity + competitive base. Specific numbers in conversation.

03

Solutions Architect — Finance / FinTech

Front-end of our finance-vertical engagements. Discovery conversations with brokerages, exchanges, fintech operators. Translate internal-analytics, compliance, or operational pain into proposals our engineering team can actually scope and ship.

You’ve already done

  • Worked inside or sold into a brokerage, exchange, or fintech platform.
  • Read a 10-K cover to cover and could explain the revenue model in two sentences.
  • Know what FIX, MNPI, best execution, and AML actually mean in practice — not just as acronyms.

You’re not the right fit if

You're an “AI strategy” person without engineering background. We need someone who can scope and propose, not advise.

Compensation

Equity + base + variable on deals closed.

How to apply

Email careers@datalabz.pro with:

  • A two-line summary of who you are.
  • A link to something you shipped (code, system, product) that you owned end-to-end.
  • A two-line summary of what role you’d want and why.

No CVs in the first email. We’ll ask if we want one. We reply within 5 business days, including to no-thanks.

Section 5

Contact

We reply within 2 business days, including to no-thanks.

Email us → Book a 30-min call

If you’re hiring us, write a few lines about the problem and what would make this work successful. If you’re applying for a role, see Careers above. If you’re a vendor, please don’t — we don’t take cold sales pitches.