Mario Erazo

Line 1

Solution architecture and integration

Platforms that stay simple while the number of clients, systems and teams grows.

1. Platform design

   Platforms on cloud infrastructure and container orchestration, designed so that client-specific customisation extends the shared product instead of fragmenting it.

2. Integration

   Integration layers across synchronous APIs, messaging and file-based exchange.

   In practiceClient and partner systems that each speak a different protocol connect to one commerce platform without fragmenting the shared product.

3. Domain modelling

   Commerce and marketplace scenarios with multiple sourcing, delivery and fulfilment options.

4. Cost management

   Infrastructure and tooling kept in line with what the platform needs.

   In practiceCloud bills that grow with every new tool shrink once infrastructure is consolidated and heavy tooling gives way to leaner alternatives.

Line 2

Engineering leadership

Technical direction and delivery for distributed teams.

1. Development standards

   Technical standards and review processes.

2. Platform simplification

   Components and technical standards unified across the platform.

   In practiceA platform that has grown service by service becomes simpler to operate once components and technical standards are unified.

3. Delivery organisation

   Operations, development and quality assurance led as one organisation, with requirements, initiatives and compliance obligations turned into prioritised work packages and the client relationship managed.

   In practiceDistributed teams that pull in different directions become one delivery organisation with shared standards and clear priorities.

4. Release cadence

   Interfaces and delivery sequencing coordinated with other teams and contractors.

   In practiceSlow, unpredictable release cycles become short and regular once delivery processes are aligned across teams.

Line 3

AI-assisted engineering

Codebases and pipelines that agents can work in safely, and teams that know how to use them.

1. Agent-ready codebases

   Standardised repository structure, co-located dependent code, and architecture and agent instruction files at repository and module level.

   In practiceChanges that stall on missing context reach delivery faster once repositories and their documentation are structured for engineers and agents alike.

2. Context engineering

   Shared tooling that gives agents the company's context.

   In practiceAgents that know nothing about a company's systems receive the context of every repository, its relations and the company's rules through shared Claude Code tooling.

3. Guardrails

   Specifications and review gates for AI-assisted development.

   In practiceBuilt this website from one specification that several AI models implement independently, so that their results can be compared directly.

4. Team enablement

   The practices of AI-assisted development introduced to the team.

• Content scattered across separate systems becomes findable through one federated search.
• Product data from many sources becomes searchable on a search product shared by several clients.
• Shop events turn into relevant recommendations through collaborative filtering.
• Microservices run reliably on container platforms with automated delivery pipelines.
• AWS Certified Solutions Architect – Associate

1. Context first

   I gather the business and technical context first, investigate the options and weigh their trade-offs. I write an RFC after consulting the developers who will build the solution, and record the outcome in an architecture decision record.

2. Peer leadership

   I lead as a peer, support engineers in doing their best work, and keep work progressing. I change my position when a different opinion rests on a real problem that the other person has understood.

3. People development

   Delegation is part of planning. The development of each engineer follows two inputs: the engineer's own goals, and what the current requirements need.

4. Shared product

   A client requirement first passes a business analysis of its relevance. A relevant requirement is generalised into a product feature. A requirement that cannot be generalised is funded by the client and released behind a feature flag for that client only.

5. Simplicity

   Every service, database and pipeline adds operating cost and cognitive load. I treat consolidation as a deliverable and plan it like any other feature.

6. Documentation

   Architecture notes and agent instructions live in each repository and, where needed, in each module. Engineers and agents see what a change affects before they make it, which keeps changes targeted.

7. Measured outcomes

   I judge a change by its measured effect on cost, lead time and release cadence, compared before and after the change.

8. Quality assurance

   Quality is part of delivery, not a phase after it. I lead quality assurance alongside operations and development, so that a change is tested and reviewed before it is released.

9. Compliance

   Compliance obligations enter the backlog like any other requirement. I translate them into prioritised work packages together with client requirements and product initiatives, so that they are planned and verified rather than discovered late.

Since 2017, building and leading commerce platforms, from recommendation and search engineering through product ownership to solution architecture and the leadership of distributed teams.

1. 2017 – 2019 Software Developer AOE GmbH Recommendation engine and core microservices
2. 2019 – 2021 Technical Product Owner, Search and Recommendation AOE GmbH, Wiesbaden Search product for external clients and internal products
3. 2021 – 2022 Technical Product Owner Cloudflight GmbH, Cologne Logistics software in a multi-contractor SAFe programme
4. 2022 – 2025 Solution Architect Omnevo GmbH Integration architecture and cloud cost for a commerce platform
5. 2025 – present Lead Engineer Omnevo GmbH, Wiesbaden (remote) Operations, development and QA teams; platform simplification; AI-assisted engineering