Paweł Procaj
Principal Engineer · AI & Data-Intensive Platforms
I design and deliver AI-native platforms and the data-intensive systems underneath them — declarative workflow models, metadata catalogs, and AWS backends — and stay hands-on in the code.
Best fit where several teams need one governed delivery model that now has to account for AI-driven workflows, and where architecture ownership still means writing and reviewing production code.
20+ years building
- ▪Enterprise software & SaaS platforms
- ▪Internal developer platforms (IDP)
- ▪Data-intensive systems at scale
Domains: Banking · media · retail · aviation · technology
Current focus
- ▪AI platform engineering: workflow systems, model integration, governed agent access
- ▪Metadata-driven execution platforms (Airflow, Spark, CI governance)
- ▪AWS data and API platforms (Python, Node/TypeScript)
Open to
- ▪Hands-on Technical Lead / Principal Engineer roles
- ▪AI-native product teams where architecture and code review sit with one senior engineer
- ▪B2B / contract engagements, remote Poland / EU
Featured · platform case study
Oct 2022 – Jun 2026MEF — Metadata Execution FrameworkNatWest (via Accenture)
A declarative platform for regulated data delivery — one workflow model, generated Airflow/Spark artifacts, and governance enforced in CI.
What I build
Governed AI access to platforms
Expose platform capabilities to assistants through controlled APIs and tools — capability discovery, not ad-hoc scripts.
Internal developer platforms
Declarative models that generate Airflow/Spark artifacts, move standards into CI, and give delivery teams a shared path to production.
Metadata & data platforms on AWS
Catalogs, ingestion, serverless APIs, and the wiring between warehouse estates and the teams that use them.
Hands-on technical leadership
Design the system, write the hard parts, own code review and workshops so adoption sticks beyond the first team.
Selected work
AI Video Platform — Image Composition WorkflowLemonlight (US)
Aug – Sep 2026Client engagement under ProLogic Labs: designed and delivered a layered image-composition workflow into a production AI video platform's TypeScript monorepo.
Delivered the contracted scope in full — architecture, integration, and handoff — inside a previously unfamiliar production codebase.
Outcome
Delivered the contracted scope in full — feature, experiment, improvement plan, and written summary — plus architecture and handoff documentation
MEF — Metadata Execution FrameworkNatWest (via Accenture)
Oct 2022 – Jun 2026A declarative platform for regulated data delivery — one workflow model, generated Airflow/Spark artifacts, and governance enforced in CI.
Replaced team-specific workflow implementations with one governed platform model for delivery across multiple bank teams.
Outcome
Designed and led the declarative execution framework used as the shared delivery model across multiple bank teams
Metadata CatalogNike
Oct 2020 – Sep 2022Graph-backed metadata catalog for Nike's data assets — improving discovery and shared context across Hive and Snowflake.
Outcome
Built graph-backed catalog APIs and metadata platform components for internal data discovery
Serverless data & API platformsThomson Reuters
Oct 2017 – Sep 2020Three years shipping production Node.js/TypeScript serverless APIs and data backends on AWS.
Outcome
Built and shipped production serverless APIs in Node.js/TypeScript on AWS
Hero prototype — AI video pre-productionProLogic Labs
2026 · live prototypeClickable prototype of a human-in-the-loop creative workflow: brief → concept → script beats → storyboard → review.
Built to demonstrate how AI-assisted video creation should work in production: versioned artifacts, human review checkpoints, and reusable workflow objects rather than one-shot generation.
Outcome
Built solo in a few days and shipped as a live clickable prototype
Flight Operations System · Crew ManagementCasses → Smart4Aviation
Apr 2003 – Dec 2006Worked in a ~12-person product team building a Flight Operations System for real airlines, with hands-on work in the Crew Management subsystem.
Outcome
~12-person product team across engineering, QA, frontend, admin, and ops
BudoArenaProLogic Labs
2025 – ongoingMartial-arts media platform grounded in 20+ years of practice and coaching — the real-world sandbox for authoring-to-video workflow ideas.
Outcome
Proves the creative/media focus is grounded in a lived domain, not opportunistic
Foundations · 2000–2017
Two decades of the same thread — platforms other engineers build on: from C++ network management, through enterprise Java and seven years at Intel (DevOps toolchain, then the open-source Trusted Analytics Platform), to banking-scale Big Data.
Contact & links
Gdańsk, Poland · Open to hands-on Technical Lead and Principal Engineer roles on data-intensive platforms — remote (Poland / EU), B2B