Alber.

Eight technical plates.Two decades of real systems.

A selection of projects that represent the kind of work I do: enterprise integration at scale, architecture of complex systems, predictive analytics and agents operating inside the business.

  • Cloud brokering with mutual SSL

    Pioneer architecture for message brokering in cloud containers with SSL and mutual authentication, at a time when this approach was rare. Goal: minimise hardware and licensing costs without compromising security. The pattern I designed would take years to normalise as the modern API gateway.

    1. Applications
    2. Message brokercloud containers
    3. Services
    Simplified flow.
    ContainersSSL/mTLSMessage brokerCloud
  • NXT: the chain as the integration channel

    NXT launched in 2013 as one of the first proof-of-stake blockchains, and one of the first to put messaging and other business functions straight into the protocol. I built proofs of concept for B2B integration that used the chain's own messaging layer as the channel between companies: signed, timestamped messages replicated across every node, with no central broker. A radically different approach from the classic middleware I was deploying at the time, and the line of work that later led me to Jelurida, the team behind NXT and Ardor.

    1. Company A
    2. NXT chainnative messaging, no broker
    3. Company B
    Simplified flow.
    NXTProof of stakeOn-chain messagingHTTP APIB2B
  • Correos ↔ Amazon

    Design, development and delivery of the full integration between Spain's postal service and Amazon. B2B solution with secure HTTP REST/SOAP over WMB. Interoperability between XML, EDI/Edifact and proprietary formats. MQ network for event distribution. DatacaptureStore tool for visualisation, exception handling and message replay. A mapping error here means packages lost at national scale.

    1. Amazon
    2. B2B integrationWMB: XML, EDI/Edifact
    3. MQ networkevent distribution
    4. Correos
    Simplified flow.
    WMBMQHTTP RESTSOAPEDI/EdifactXMLIHSTDS
  • Madrid Smart City: MiNT Protocol

    I defined every external interface for the City of Madrid's smart-city initiative. Hybrid cloud architecture, full security (SSO, non-repudiation, integrity, encryption) and integration patterns across JMS, HTTP-JMS REST, SOAP web services and MQ front-ends. Integration with IBM MAXIMO and IOC for a city of 3 million people. Led and coached the middleware team.

    1. External interfaces
    2. SecuritySSO, non-repudiation, encryption
    3. Integration busIIB, MQ, DataPower
    4. MAXIMO and IOCIBM
    Simplified flow.
    IBM IIBMQDataPowerMAXIMOIOCSSOHybrid cloud
  • Severn Trent Water: Predictive analytics

    Real-time analytics on the UK water distribution grid. Multi-sensor collection (digital, analog, weather) through a SOAP-MQ gateway in DataPower. ODM rules engine orchestrated from IIB v10. The system detected incidents before they happened. Also: corporate address-management API with ESRI and GIS integration, SQL Server to Oracle migration, and CI with Jenkins.

    1. Sensorsdigital, analog, weather
    2. DataPowergateway
    3. IIB v10
    4. ODM rules
    5. Incident alert
    Simplified flow.
    DataPower v7ODMIIB v10MQ v8SQL ServerESRI/GISJenkins
  • Senator Project (H2020)

    Technical leadership within the Senator initiative, funded by the European Union under the Horizon 2020 programme. Advanced proofs of concept, official technical documentation for the Ardor ecosystem, and tutorials for enterprise adoption. Four years of international collaboration from Jelurida, the team behind NXT and Ardor.

    JavaArdor PlatformREST APIEU H2020
  • PradoMusicAcademy: end-to-end agents

    AI agents running the complete operation of a music academy. Architecture with state, inter-agent orchestration, integration with business APIs, observability. Result: cut from 13 hours to 2 hours of daily manual management. The director got back time for the creative side of the business.

    1. Daily operationsemail, payments, rosters
    2. Stateful agentsorchestration and traces
    3. Business APIs
    Simplified flow.
    AgentsOrchestrationLLM APIsAutomation
  • Restaurant group: front-desk and booking agent

    Conversational agent embedded in the website of a Madrid restaurant group. Its own persona in the house's tone, bookings written straight to the venue's database, daily and weekly reports to management over Telegram and conversion analytics for every conversation. From pilot to production without changing the architecture: the model gets swapped, the integration and the traces stay.

    1. Website chat
    2. Agent with a personathe house tone
    3. Bookings DBthe venue's own
    4. Reports to managementTelegram
    Simplified flow.
    Conversational agentPersonaBookings DBTelegramAnalytics

Teaching and transfer.

IBM Educational Consulting and Draper University

Official IBM technology courses across Europe. Clients: BEC.dk, central banks of Denmark and Spain, Asisa, Accenture. WMB v6 to v8 migration training in Saudi Arabia. In 2023, Draper Hero Training in Silicon Valley, AI hackathon at Google and a generative-AI event at OpenAI.

WMBIIBMQDataPowerSOAGenerative AI

Does any of thislook like what you're dealing with?

One hour to review your architecture, your AI plan, your critical integration, with someone who has been building these systems since 2002.

Book a session (€500)