Berlin, Germany

Senior Java Backend Engineer
& Private AI Systems

20+ years building reliable backend systems with Java and Spring Boot — with additional expertise in AI, RAG, local LLMs and on-premise deployment.

Available for freelance and contract projects.

View Services Recent Work

Freelance Services

Backend engineering first. AI where it adds value.

I help organizations design, build and extend reliable software systems — from Java/Spring Boot backends and APIs to practical private AI solutions. The focus is production-oriented engineering, integration and delivery.

Java Backend Engineering

Senior development for enterprise and distributed systems: Java, Spring Boot, REST APIs, Kafka, PostgreSQL, persistence, testing, Docker and system integration.

AI & Knowledge Systems

RAG, semantic and vector search, document intelligence, knowledge graphs and LLM integration — applied as part of a robust software architecture.

Explore AI capabilities →

On-Premise AI & FDE

Forward-Deployed Engineering (FDE) for private AI: install, configure, integrate and adapt local LLM systems on customer infrastructure where data must remain under organizational control.

AI & Private AI

AI integrated into real software systems.

I build and integrate AI components where they solve a concrete information or workflow problem — including local models and on-premise systems for organizations that need control over their data and infrastructure.

Information Retrieval

Relevant information across large, heterogeneous document collections using semantic retrieval, vector search, lexical matching, reranking and query-time filtering.

Knowledge Representation

Knowledge graphs, entity relationships, metadata and graph-based reasoning that make organizational information explicit and queryable.

Document Intelligence

Parsing, classification, chunking and embedding pipelines that prepare documents for retrieval, verification and LLM-assisted workflows.

On-Premise & Private AI

Deployment and adaptation of local AI systems using local LLMs, GPU infrastructure, Docker and APIs — designed around the customer's existing environment.

Solutions

Current work.

Selected recent work demonstrates how I apply the same backend engineering principles to AI-intensive applications, including retrieval, verification, document workflows and local language models.

Java 21Spring Boot 3.3PostgreSQLQdrantNeo4jOllamaRAGKafkaOn-Premise AI
2026 — Current

Verwaltungsassistent — Municipal AI Decision Support

An open-source system for case-based decision support in public administration. It ingests citizen e-mails and municipal documents, matches information to cases, retrieves the relevant evidence and produces traceable decision support. Claims are verified against the retrieved evidence; when the evidence is insufficient, the system fails closed instead of guessing.

The project combines a Java/Spring Boot application with semantic retrieval, deterministic rule evaluation, verification, audit trails and locally hosted language models. A dedicated AI showcase is available at Maxwell-Nord.

Java 21Spring Boot 3.3PostgreSQLQdrantNeo4jOllamaQwen2.5:14Bnomic-embed-text
AI showcase & live demo → View source on GitHub →

Live demo: Monday–Friday, 12:00–17:00. The demonstration system is intentionally offline outside these hours.

Verwaltungsassistent dashboard

Professional Experience

Over twenty years of software engineering.

2024–Present

Independent Software Engineering & AI Systems

Berlin / Remote

Designing and implementing backend and AI systems with a focus on knowledge-intensive applications, retrieval-augmented generation and private/on-premise deployment. Current work includes the open-source Verwaltungsassistent for municipal casework, local LLM integration, document intelligence, semantic retrieval and verifiable AI-assisted decision support.

Java 21, Spring Boot 3.3, PostgreSQL, Qdrant, Neo4j, Ollama, RAG, Docker, Local LLMs

2022–2023

T-Systems International

Frankfurt am Main (remote)

Wrote software for Operations Control Centers used by German police, fire departments and emergency services — real-time, fail-safe systems.

Java 17, Kafka, SOAP, REST, Vert.x, WebSockets, Maven, Liquibase

2021–2022

Inverso GmbH

Munich (remote)

Subsidiary of VKB (Versicherungskammer Bayern). Developed software components in a Kafka pipeline and Spring Boot-based services.

Gradle, Kafka, Spring Boot, JUnit, Java 11

2021

Dønvold Detour ENK

Norway

Built a Spring Boot backend with user management, secure login, OpenAPI v3.0 compliant REST endpoints, Postgres database and Docker deployment.

Docker, Maven, Spring Boot, JUnit, Mockito, Postgres, OpenAPI, REST, Java 8, Lombok

2019

Ingrano Solutions GmbH

Berlin

Developed automated tests for router software as part of the digitalization of the German healthcare system (telematics infrastructure).

Docker, Maven, Spring Boot, Cucumber, JUnit, Java 8

2013–2019

Private Client

Switzerland

Long-term project for time-series prediction using classification, regression and deep learning. Applied WEKA and DL4J for data mining, neural nets, convolutional nets, transfer learning and LSTM networks. Later rewrote the application as microservices with Spring Boot, adding an AlphaGoZero-based service with Kafka and WebSockets.

Java 8, Maven, WEKA, DL4J, Docker, Spring Boot, Zuul, Cassandra, Scala, Kafka

2017

German Aerospace Centre (DLR)

Berlin

Mandatory university internship at the DLR Institute of Planetary Research. Processed seismic data from the ROBEX lunar rover prototype.

Java 8, Python

DLR German Aerospace Centre
2015–2016

Bonify

Berlin

Credit report startup. Developed microservices for the backend in a small team.

Java 8, Maven, Postgres, JavaMail, SOAP, Flyway, Mockito, JUnit, Spring Boot

2012–2013

Fraunhofer FOKUS

Berlin

Fraunhofer Institute for Open Communication Systems. Worked on civic technology portals including Fix-My-City and Adhocracy, integrating LiquidFeedback with CKAN and Spring Boot.

Java 8, Maven, Spring MVC, JUnit, Hibernate, Postgres, Lua Server Pages

Fraunhofer FOKUS
2009–2011

Max-Planck-Institut für Bildungsforschung

Berlin

EU-funded LarKC project for massively distributed incomplete reasoning. Implemented data mining applications using clustering and latent semantic analysis. Worked with Semantic Web technologies.

Maven, Spring MVC, LingPipe, Hibernate, Postgres, OpenNLP, SemanticVectors, Sesame, R

Max-Planck-Institut für Bildungsforschung
2009

DFKI

Berlin

German Research Center for Artificial Intelligence. Worked on a Sentiment Analyzer project — implemented software for clustering, classification and social media mining.

Java, WEKA, LingPipe

DFKI Berlin
2008

CMS, Humboldt University of Berlin

Berlin

Integrated Alfresco CMS with Liferay portal for the Computer and Media Service Department.

Java, CAS, LDAP, Alfresco, Liferay

2007

Nokia Siemens Networks

Berlin

Joint venture between Nokia Oyj and Siemens AG. Developed modules for the internal Nokia-Siemens portal.

Java, JSP, Spring Framework, Hibernate, JPA, Oracle

Nokia Siemens Networks
2005–2007

Siemens Networks

Berlin

Communications and information business of Siemens AG. Developed modules for the internal Siemens portal used by managers worldwide.

Java, JSP, Spring Framework, Hibernate, JPA, Oracle

Siemens
2005

European Patent Office

Berlin

Analyzed patent documents in the field of Natural Language Processing and defined proposed new classification categories.

EPO internal portal, US Patent Office portal, Japanese Patent Office portal

European Patent Office Berlin
2002–2004

Computer Center, Humboldt University of Berlin

Berlin

Developed a JSP-based application for the media center at Humboldt University.

Servlets, JSP, Hibernate, JPA, Tomcat, Postgres

Humboldt University
2001–2002

Tonxx GmbH

Berlin

Developed a spam-detection plugin for Microsoft Outlook based on a patent application. GUI design and implementation of the Outlook plugin.

.NET (C#)

2000–2001

Semantic Edge

Berlin

Developed natural language dialog systems. Java development, data mining modules and GUI programming.

Java, Perl

Communities

Startup Weekends.

Regular participation in startup weekends and innovation events across Europe. Built prototypes, collaborated with diverse teams and explored early-stage product ideas.

Berlin, Oslo, Bergen, Zurich, Prague, St. Gallen — 2013–2019

About

Ralph Brandão Vidal

Berlin, Germany

Over twenty years of software engineering, with work spanning Java backend development, natural language processing, information retrieval, knowledge representation, enterprise software and machine learning.

My focus is building reliable backend and knowledge systems — and increasingly deploying AI where data must remain under organizational control. This has led from early work in content management and enterprise portals through semantic search and knowledge graphs to RAG, local language models and on-premise/private AI systems. Machine learning and language models are components of these systems — not the systems themselves.

Advanced studies (Hauptstudium) in Computer Science at Humboldt University of Berlin. M.Sc. in Geological Sciences (Geophysics), Freie Universität Berlin. Native Portuguese speaker. Fluent in German and English.

Java Spring Boot Neo4j Qdrant (vector store) PostgreSQL Kafka Docker REST APIs LLMs RAG On-Premise AI

Contact

Get in touch.

LinkedIn
linkedin.com/in/ralphbrandaovidal
Phone
+49 1573 020 6486
Email
contact@brandao-vidal.com
Location
Berlin, Germany

Open Source Projects

Public repositories.

A selection of public repositories, experiments and technical demonstrations related to enterprise software, geospatial systems, machine learning and information retrieval.

01

Verwaltungsassistent

Open-source municipal AI decision-support system combining case workflows, document-grounded retrieval, deterministic verification and local LLMs.

Java 21 · Spring Boot 3.3 · Qdrant · Neo4j · Ollama

AI showcase & demo → View on GitHub →
02

Hydro Sentinel

Real-time flood monitoring and alert simulation platform with multi-view dashboards, staged alert escalation and dynamic water-level propagation.

Java 21 · Spring Boot 3.3 · React 19 · TypeScript · PostgreSQL

View on GitHub →
03

Geospatial Search with PostGIS

Spatial search using Spring Boot, Hibernate Spatial and PostGIS, with REST endpoints for proximity queries.

Java 8 · Spring Boot 2.5 · PostGIS · Hibernate Spatial · Docker

View project →
04

Satellite Image Classification

Transfer learning with DL4J and VGG16 for satellite imagery classification.

Java · DL4J · VGG16 · Transfer Learning · Computer Vision

View project →
GitHub
github.com/rbvidal
GitLab
gitlab.com/rb2000

Why Knowledge Systems Matter

The problem is not a lack of information.

Organizations already possess enormous amounts of valuable knowledge — technical documentation, policies, contracts, engineering specifications, research reports and internal procedures. The challenge is that this information is scattered across multiple systems and difficult to retrieve when needed.

Traditional keyword search solves part of the problem: it finds documents containing specific terms. But it does not understand meaning. It cannot connect related concepts across documents, and it cannot reason about the information it retrieves.

Modern knowledge systems take a different approach. They combine semantic retrieval, vector search, knowledge graphs and language models to help people find, understand and use organizational knowledge more effectively.

Large language models work best when they augment existing software — not when they replace it. The strongest results come from systems where language models are one component of a broader information architecture, not an isolated interface.

Organizations are increasingly evaluating private and open-source models that run inside their own infrastructure. This keeps sensitive information under control while still benefiting from modern capabilities.

The objective is not simply to use AI. The objective is to improve how organizations capture, retrieve, connect and operationalize knowledge.

The platform below demonstrates how these concepts combine into a practical software system for document intelligence, semantic retrieval, workflow automation and knowledge-driven applications.