/engineering-lab // where production happens
The Engineering Lab is where we run our own company through the same machine we propose to clients. Every library, every RPA, every AI agent, every integration — verified internally first, only then offered externally. Echo of ten LSO:ATOM protocols: every delivery verifiable in data, every error with post-mortem, every architectural decision with documented alternatives. 31 developers in 2 teams on competitive exclusivity — complete operational discipline, not developer rental.
Engineering Lab Products // target sector
The Engineering Lab does not push generic SaaS. Each product is a proprietary platform built around a specific sector — with LSO:ATOM protocols embedded from day one.
Sector // higher education
AI platform for universities — efficient dean's office, better teaching, and hybrid classes in one environment.
Service // websites AVAILABLE
From business card to platform — designed to convert
We design and build websites for SMBs: from simple landing pages to full platforms with CMS and integrations. Every site: fast, mobile, WCAG-compliant — and designed to turn visitors into clients.
In development
More AI platforms for regulated sectors — active conversations with industry partners. Ask about pilot availability.
Specialization areas // 6 capabilities
Connecting ERP, CRM, WMS, and industry-specific systems into a single coherent data layer. Real-time integration — systems exchange data continuously, no overnight syncs. Typical case: 4-8 systems that do not talk to each other. Stack: REST + OpenAPI 3.1, GraphQL, webhooks; integration platforms (Make/n8n/Workato) for SMBs; dedicated middleware (.NET / Node) for security requirements. Advanced API security in every deployment.
Data flows from source to report — production data transfers between systems that do not crash in the middle of the night. Orchestration (Airflow, Prefect), transformation (dbt), loading (Snowflake, BigQuery, Databricks for Spark). Scattered Excel and SQL data consolidated into one place. Legacy system migration to cloud with rollback capability.
RPA, workflows, and API integrations for processes too complex for Power Automate and too simple for a full ESB. UiPath, Power Automate, Make/n8n, or Python with FastAPI as trigger endpoints. Document AI (Azure Document Intelligence, Google Document AI) for ingesting invoices, orders, certificates. Typical case: a series of manual steps between systems (copy-paste, validate, report) consuming hours daily. Deployment 4-16 weeks with 12-month post-deployment support.
Production applications where off-the-shelf solutions fall short. Full web stack (Astro/Next.js/React), backend (Node/Python/.NET), databases (PostgreSQL/SQL Server/Snowflake) + Redis as cache. Typical case: unique processes, scale too small for enterprise vendor, or regulations excluding public cloud. Performance guarantee with contractual penalty — written into every QDeployment contract.
Seeing what happens in production systems before the client calls. Grafana + Prometheus for metrics, Sentry for application errors, Datadog/New Relic for full tracing, ELK stack (Elasticsearch + Logstash + Kibana) for logs, dedicated business dashboards. The technical team learns about outages from the system, not from users.
Anomaly detection in operational data, demand forecasting, churn prediction. MLOps (MLflow / Vertex AI), automatic model retraining, data drift monitoring. Advanced security (NIST SP 800-207), NIS-2, GDPR. Compliance documentation for regulated clients (energy, finance, healthcare).
Engineering principles // 3 of 10 LSO:ATOM atoms
ATOM 01
Every process we implement has a finite number of states. Every transition is logged. No "magic black box" — everything is traceable from trigger to effect.
In practice: every workflow in QDeployment generates an audit trail in QCare telemetry. Production incident -> root cause in <30 min.
ATOM 03
Every operation can happen again without side effects. Every operation can wait. Communication between systems via queues (RabbitMQ / Kafka), never via synchronous API that blocks.
In practice: external API outage does not equal client system outage. Operations retry until success, the queue grows but never loses data.
ATOM 09
Every process has a `.md` in the client repo. Updated on every deploy. No "documentation somewhere in SharePoint". The runbook lives in git, peer-reviewed like code.
Tech stack // production-verified
Implementations // micro-excerpts from case studies
Collaboration process // 5 steps, each with an artifact
Five steps, each with concrete artifacts the client receives in hand. No "PowerPoint reports", no vendor lock-in, no surprises at the end.
First 1-3 weeks depending on project scale. We understand the technical context (which systems exist, which do not, infrastructure constraints) and business context (what must work, for whom, by when).
Build vs buy decision per component, C4 diagrams (context, containers, components, code), trade-off analysis between variants. We document enough that the client and their technical team can evaluate decisions and understand a year later why we chose this architecture.
Two-week sprints with clear artifacts at the end. Kanban for continuous projects, Scrum for fixed deadlines. Your team has repository access from the first commit.
Automated tests (unit, integration, end-to-end), staged rollout (dev -> staging -> production), monitoring configured before go-live, not after.
Your team takes over the project and documentation, we remain in the background as second-line support for the first 4-12 weeks post-deployment (depending on scale). The goal is not client lock-in, but that after we leave the system operates without us.
FAQ // 9 questions we answer in the first call
Default T&M — in engineering projects it is hard to predict scope with accuracy required for sensible fixed-price. For well-defined projects up to 4 weeks we offer fixed-price. For longer projects we work T&M with a weekly budget cap and weekly reporting.
The client. Standard clause in our contract: all code written for a client project is client property from the moment of creation. Open-source code remains under original licenses.
Yours. We prefer committing from day one to the client repository (GitHub, GitLab, Bitbucket, Azure DevOps). If the client does not yet have a repository, we set one up on their account.
Every pull request goes through review by at least one engineer from our team before merge. Standards: PR with description, tests for new functionality, no linter warnings, passed CI/CD pipeline.
Simple middleware integrating two systems: 3-6 weeks. Data pipeline with warehouse and dashboards: 6-12 weeks. Custom business application: 3-6 months. Projects over 6 months are split into phases with separate checkpoints.
Yes. Your team has repository access from the first commit. Demo at the end of every 2-week sprint (~30 min Teams/Meet). Weekly email update with list of completed work and plan for next week.
We start with discovery, not a contract for the whole project. 2-4 weeks discovery for 15,000-30,000 PLN, at the end the client receives a Technical Discovery Report, architecture recommendation, and estimate for the full project.
We prefer Python (back-end, data), TypeScript (frontend, Node.js backend) and PostgreSQL (database). For infra AWS and Kubernetes. But your stack may differ — we adapt, we do not change it against client realities.
If it is a niche database (Cassandra, Neo4j), we can typically ramp up in 2-4 weeks. If completely exotic technology (mainframe COBOL, niche SAP module, hardware firmware) — we honestly recommend someone who will do it better.
Technical consultation // directly with dev / data lead
If you have a specific engineering project to discuss — or a concept that needs technical validation before a decision — start with a 30-minute call. We do not sell on the first call. Either we see that we can build what you need, or we do not — and in the latter case we will say so directly.
Book a technical consultationTechnical consultations are led by Antoni Draminski (Head of Data) or Mateusz Jakimow (CEO). 30 min, no commitment.