AI Consulting & Mentoring

I build the agentic layer —and teach your team to run it.

Senior full-stack engineer turned agentic-AI specialist. I architect durable, self-hosted multi-agent systems with real operational controls — budget, concurrency, cost-tracking, human-in-the-loop — then hand them to your developers under audit and governance.

Let's talk→What I do

10+ yrs in productionavailable for engagementsin cooperation with

autofix.runrunning

agent · autofix

// what i do

I build the agentic layer. And I teach your team to run it.

I don't show up with a tool and leave. I set the context AI works in reliably — instructions, boundaries, skills, MCP, measurement — wire it into your stack, and hand it to the people who'll keep working in it long after I'm gone.

01context-engineering

Context engineering

Moving from prompting to running context is a craft, not a setting. I teach the team to write instructions and AGENTS.md, hold boundaries, give the agent memory and feedback from real code. The result doesn't depend on a specific model or tool — it survives the next model swap.

AGENTS.md · skills · memory · boundaries

→Instructions, AGENTS.md and shared memory across the team
→Boundaries: AI never touches auth, payments or keys
→Processes independent of any single model or tool

02agentic-layer

Agentic layer in production

I turn repeated work into a skill or sub-agent that does it the same way every time — PR descriptions, test generation, code review, cross-codebase research. It doesn't stop at a demo: the layer ships into your workflow and runs in production with budget, concurrency and cost-tracking.

skills · subagents · cost-tracking · HITL

→Skills for repeatable tasks in your stack
→Sub-agents & parallel research, verified against code
→Runs in production — not a one-day demo

03autofix

Observability → autofix

I wire an agent to your observability. A new error lands — the agent reads the stack trace, code and tests, reproduces it, finds the cause, writes the fix and tests, and opens a pull request. You approve it. Same principle for logging and debugging: a shorter path from incident to fix.

sentry.issue → reads context → fix + tests → PR

→Reproduce, fix and test from a real stack trace
→A PR to review, not auto-merge — humans stay in control
→Same approach for logging and debugging

04mcp

Custom MCP servers

I build the MCP server through which AI understands and operates your own software — your database, internal APIs, Jira, Confluence. Deterministically, with permissions and boundaries, not screen-scraping and hoping.

mcp: db · jira · confluence · internal API

→AI ↔ your apps, data and internal sources
→Deterministic, fast, with clear boundaries
→Permissions and auditability from the first commit

05mentoring

Workshops & mentoring

Live, in your stack, I show how to run AI — tips, tricks and gotchas from real deployments, not slides. I grow a champion network that keeps the practice alive after I leave, and cover the AI-literacy duty under EU AI Act article 4.

live PoC · champion network · EU AI Act art. 4

→Live PoC in your own code, not a generic demo
→A champion network that holds the practice
→AI-literacy per EU AI Act, article 4

// the foundation

Safe, governed adoption.

Under the whole layer sits governance — DORA, EU AI Act, GDPR. AI never touches auth, payments or keys. I measure what matters: cycle time, change-failure rate, time from incident to fix. No hero numbers.

DORA · EU AI Act · GDPR · boundaries over auth, payments, keys

// loop engineering

I build the system that runs the agents.

Loop engineering: instead of prompting AI by hand, I build the durable loop it runs in — observed signal in, reviewed pull request out, a governance gate before anything ships.

AI doesn't write instead of your team. It works inside a context that has boundaries, memory and measurement.

autonomy: L1 → L3, rolled out, not flipped
human-in-the-loop: gates where it matters
self-hosted: your data, not someone's cloud
audited: every action logged & reversible

// selected work

Production agentic systems — built, shipped, governed.

A selection of the systems behind the practice. Own products, work built at ADF, and enablement for enterprise clients — described honestly, no invented metrics.

eve.dev.lovinka.com

evelive

7 agents · survives SIGKILL · 38-tool HITL control plane

Architect·built at ADF

eve

A self-hosted agent platform that survives a hard kill — and resumes the human approval it was waiting on.

Seven agents — a root orchestrator plus migration, PR-review, Sentry-fix and outreach specialists — run on a Postgres-backed workflow world. A run paused mid-flight on a human-approval gate survives a hard SIGKILL and resumes on restart, with idempotent side effects. A live 38-tool control plane sets approval policy per tool; a React 19 observability console sits on top. No third party in the data path — LiteLLM routes to native Anthropic, native OpenAI and a self-made Codex-pool gateway.

React 19
Node 24
Postgres
LiteLLM
native Anthropic
Tauri / Rust

eve.dev.lovinka.com

reservine.io

Reservinelive

Own Angular UI library (73 components) · 19-provider AI backend

Founder · Frontend Architect·own product · co-founded

Reservine

Founder-built booking SaaS with a deep, multi-provider AI backend.

A multi-tenant reservation platform for any business. I founded it and own the Angular 21 Signals frontend and its 73-component UI library; the product runs 16+ AI services on a 19-provider LLM abstraction with structured output. Built with co-founder Martin Foltýn, who owns the Laravel backend.

Angular 21
own UI library
Nx
Laravel 12
multi-tenant
Stripe Connect

reservine.io

fixit.app

FixItlive

Governed in-product AI · tri-stack RN / NestJS / Go

Founder · Architect·own product · LEFTEQ

FixIt

Real-time services marketplace with an in-product AI layer.

An Expo / React Native + NestJS + Go marketplace — recognise a problem from a photo, live tracking, in-app payments. Its in-product AI layer ships with real operational controls (budget, concurrency, idempotency, cost-tracking) and a Codex-pool gateway, on a multi-runner test pyramid (Playwright + Appium + unit/integration).

Expo / RN
NestJS
Go
PostGIS
Playwright
Appium

bpmn.app

AI BPMN authoringlive

AI output golden-tested byte-equal to the human Designer

Senior FE Architect·built at PowerFLOW

AI BPMN authoring

An agentic loop that authors BPMN — provably the same XML a human designer would draw.

At PowerFLOW I built a production AI BPMN-authoring feature: a from-scratch NestJS microservice running a server-side agentic tool-loop over SSE (function-calling + RAG), feeding edits into a real bpmn-js Camunda-7 Designer canvas. A deterministic factory turns the model's output into Camunda business objects, golden-tested byte-for-byte against the human Designer. The PoC helped win the CETIN contract.

NestJS
bpmn-js
Camunda 7
OpenRouter
RAG
Angular

deployik.app

Deployiklive

90+ tool MCP server · single-binary Go PaaS

Author·own product · LEFTEQ

Deployik

A self-hosted Go PaaS that agents can drive.

A single-binary, self-hosted platform-as-a-service in Go — blue-green deploys, auto-SSL, AES-256-GCM secrets — fronted by a TypeScript MCP server of ~90 tools so an agent can run the whole deploy / diagnose / screenshot loop in natural language. The site you're reading deploys on it.

Go
chi
go:embed
Docker
MCP (TS)
blue-green

csob.app

ČSOB Stavební pojišťovnalive

Enablement kit · eval harness · safer on regulated code

Consultant (via ADF)·external consultant — not on their dev team

ČSOB Stavební pojišťovna

Enabling a bank's developers to adopt agentic coding.

As an external consultant via ADF, I built their GitHub Copilot / AIX enablement kit — reusable skills, custom agents and prompts across ~10 repos, with banking-safety guardrails (auth / payment / key-management off-limits, anti-fabrication) and DORA-aware PR review. A reproducible eval harness — golden tasks mined from their own repos, A/B'd across models — proved the kit even makes weaker models refuse-and-escalate on risky changes instead of waving them through.

GitHub Copilot
AIX kit
eval harness
DORA-aware

lukas_pribikPrague, Czech Republic

// about

AI is a tool. The team stays yours.

For 10+ years I've built scalable, fast web applications — as a Frontend Architect and full-stack engineer, on enterprise projects and my own products. The last few years I've done something that looks like development but is a different discipline underneath: engineering the context AI works in. I strengthen teams, I don't replace them — and the same workflow I deploy for clients, I run myself every day.

10+ yrs in production · Frontend Architect / full-stack
Founder of Reservine · building FixIt
Based in Prague · in cooperation with ADF

github.com/LEFTEQ in/lukas-pribik-dev

// notes

Tips, tricks and gotchas from production.

Short, concrete notes on bringing AI into engineering without the pain — what works, what doesn't, and why. From real deployments, not talks.

All notes →

// contact

Let's map where your AI adoption has the most leverage.

A short conversation to find where an agentic layer makes the most sense — and where to steer clear. No commitment, no hype.

Write me an email→lukas.pribik@byadf.cz

or find me onGitHub·LinkedIn·+420 735 124 376