Automotive ERP
Critical ERP modernization across sales, purchasing and warehouses.
20kPrincipal Software Engineer · Legacy Modernization · ERP Architecture
I modernize critical ERP and legacy systems without stopping the business.
I take principal-level responsibility from architecture decisions to critical implementation paths and safe production delivery.
Evidence from production systems
- 20k
- orders / month
- 3M
- ERP SKU
- 40 sec
- max. legacy deployment
- No downtime
- production database migration
- 45k/min
- marketing events / minute
Results from different projects. Full context and personal responsibility are explained in the case studies.
AWS certification credential
Selected projects
How I work
Modernization built around business continuity.
A critical legacy system does not need to be rewritten all at once. First I make it safer to operate and easier to understand. Then I move complete business processes behind explicit boundaries.
- Understand the critical path
Start with the business process and production constraints.
- Modernize in reversible steps
Reduce risk through observable, incremental change.
- Own the production outcome
Keep architecture connected to delivery and operations.
Stabilize → Observe → Map → Extract → Replace
Business continuity
- Stabilize
Secure deployments, backups, access and the largest failure points.
- Observe
Add metrics, logs, alerts and business-process visibility.
- Map
Reconstruct processes, dependencies, truth sources and hidden rules.
- Extract
Move a complete process into a domain with explicit contracts.
- Replace
Switch traffic, verify the result and retire the old implementation.
Featured work · Automotive ERP
Modernizing a critical ERP without stopping sales, purchasing or warehouse operations.
A large automotive parts wholesaler depended on several undocumented, tightly coupled legacy systems. Deployments were risky, data disagreed across databases, failures were reported by users, and large supplier orders required hours of manual work.
Instead of proposing a big-bang rewrite, I designed a process-by-process modernization program. Production SCM and OMS took over core workflows, deployments became reversible, the environment moved onto managed AWS infrastructure, and legacy and modern systems gained one observability layer.
Read the full case studyVerified outcomes
- Several hours → several minutes
- for an order containing around 2,000 parts
- About one hour → max. 40 seconds
- for a reversible legacy deployment
- User report → alert within seconds
- across 14 system components and their infrastructure
- 12 GB and 59 GB databases · no downtime
- migrated from on-premise infrastructure to AWS RDS
Modernization map
- Legacy ERPProduction
- OMSProduction
- SCMProduction
- AWSProduction
- ObservabilityProduction
- PIMIn progress
- WMSIn progress
- Pricing EngineIn progress
- Marketplace IntegrationIn progress
- Central BIIn progress
When change becomes risky
Critical systems rarely fail all at once. They become slower, riskier and harder to change — until the business can no longer ignore them.
- 01
Your legacy system is too important to replace and too risky to change.
- 02
Deployments and incidents depend on one person.
- 03
Orders, products and inventory disagree across systems.
- 04
The team needs principal-level ownership, not another vendor.
Ways to engage
Start with the risk that has to be reduced first.
- 01A 90-day and 12-month Legacy Exit Map.
Legacy Modernization Assessment
A fixed-scope assessment of systems, critical processes, sources of truth and operational risk.
- 02A system that is safer to operate and change.
Stabilization & Operational Excellence
Repeatable deployments, dry-runs, health checks, rollback, observability, backup and recovery.
- 03Complete processes moved behind explicit boundaries.
ERP Architecture & Incremental Delivery
Domain boundaries, data ownership, contracts, rollout and decommissioning across ERP processes.
- 04Direct technical ownership where the risk is highest.
Principal Engineer / Architecture Partner
Architecture decisions, independent review and hands-on support without adding another management layer.
Selected outcomes
Responsibility stated as clearly as the result.
Production
One order model across critical order flows
B2B, phone, workshops and cross-dock now use one canonical model with idempotency, drafts, backorders, documents and full history.
Piotr’s responsibility: Designed and implemented end-to-end by Piotr.
Production
Supplier decisions based on price and real availability
Four supplier formats are normalized into one model; the lowest available purchase price wins and out-of-stock suppliers are excluded.
Piotr’s responsibility: Piotr designed the architecture and implemented the frontend and later changes; another developer contributed to the initial backend.
Completed
Production databases moved without downtime
12 GB and 59 GB MariaDB databases moved from on-premise infrastructure to AWS RDS after a week of observed replication.
Piotr’s responsibility: Piotr executed the migration independently.
Production
Catalog deltas processed automatically
A catalog of nearly 10 million products moved from a multi-day, developer-driven update to an automatic process triggered by new deltas.
Piotr’s responsibility: Piotr designed the architecture; other developers implemented it.
Principal-level ownership
Architecture and implementation should not be separated by a handoff.
I’m Piotr Mucha, a Principal Software Engineer and hands-on architect with experience across ERP, e-commerce, marketing automation, financial integrations, AWS platforms and applied AI.
I work directly on architecture, critical implementation paths and production risk. When more delivery capacity is needed, FairyDeck can bring in specialists without diluting my personal responsibility for the engagement.
AI-enabled engineering
AI can shorten the engineering loop. It does not replace accountability.
I use agentic tooling where versioning, review, tests and rollback keep the work auditable.
Production · 8 developers
Model-agnostic engineering CLI
GitHub remains the source of truth; rules are versioned, reversible and synchronized manually today.
Completed · 6 developers
Six-module engineering programme
Planning, subagents, skills, hooks, MCP, end-to-end validation and review for six developers starting with little prior experience.
In progress · Engineering method
Verification-driven agentic SDLC
Persistent plans, small work units, tests, browser comparison and explicit verification loops.
Product Lab
Product work keeps architecture grounded in production and R&D constraints.
Production · real users
Senioruj.pl
A silver-economy platform designed and built end-to-end by Piotr.
Active development
Home Medicine
A Flutter mobile app for medicine scanning, storage, expiry and dosing, with local-only data and QR transfer.
Active R&D · no users
Kivvi-click
An auditable product for event collection, live view, email and selected automation capabilities, built with PHP, PostgreSQL, pure JavaScript and pure CSS.
Insights
Working notes from critical-system modernization.
In preparation
Why rewriting a critical ERP is usually the wrong first move
In preparation
How a risky one-hour legacy deployment became a 40-second reversible pipeline
In preparation
Observability for business failures, not only servers
FairyDeck · Piotr Mucha
Is your critical system too important to replace and too risky to leave unchanged?
Tell me what the system does, what makes it risky and what needs to change. The first conversation establishes whether an assessment, stabilization engagement or principal-level architecture partnership is the right next step.
Discuss your systemNDA-friendly. No source-code access is required for the first conversation.