Dr.-Ing. · Senior AI Engineer

I build AI agents that work in production.

RAG, multi-agent systems, MCP and machine learning, integrated into the ERP, CRM and ticketing systems they have to work with. 15+ years of software engineering, Java and Python, Azure and AWS.

RAG · pgvectorLangGraph · A2AMCP Spring AIPython · ML
See selected work GitHub ↗
Portrait of Farid Zaid
Farid Zaid · available now
15+
years building software
4
AI / LLM systems since 2024
13
years Java · Spring
6
industries
3
languages: DE · EN · AR
What I do

From a request to a decision,
with a human in the loop where it matters.

Agentic AI & RAG

Multi-agent systems that classify, research and act through well-defined tools, grounded in your documents and data, with structured outputs, evaluation and approval steps.

  • LangGraph
  • A2A
  • MCP
  • pgvector
  • deepeval

Machine learning & data

Classification, anomaly detection and forecasting models, trained on historical data and retrained from user corrections, on Databricks and Azure Data Lake.

  • scikit-learn
  • LightGBM
  • Databricks
  • MLflow

Enterprise integration

The part that makes AI useful: REST and event-driven integration, Java/Spring Boot and Python/FastAPI services, Angular frontends, CI/CD on Azure and AWS.

  • Spring AI
  • FastAPI
  • Kafka
  • Azure
A typical system I buildrequest → agent → knowledge & tools → systems → human review
Requestemail · ticket · chat · document
→
Agentclassify · plan · act
→
Knowledge & toolsRAG · MCP · APIs · text-to-SQL
→
SystemsERP · CRM · claims · tickets
→
Human reviewconfidence · citation · approve
Live demo · built by me

Don't just read my CV. Ask it.

The chat in the corner is a production AI agent I built. It answers from my real CV and case studies, and it says so when it doesn't know. It uses the same patterns I build for clients.

RAG, grounded

CV and case studies embedded in pgvector. Only retrieved facts reach the model.

Agent Skills

SKILL.md workflows: job-fit analysis, rough estimates, architecture walk-throughs.

MCP tools

A separate MCP server for projects, availability and contact requests.

Production-ready

Streaming, rate limits, a daily cost cap, bot protection, GDPR retention.

Preview. The real agent is in the corner. Try it.

Try a recruiter's question

Each one opens the agent and asks it live. Watch its skills and tool calls as they happen.

Java 25Spring Boot 4Spring AI 2ClaudepgvectorMCPWeb Component
Selected work

Systems I've designed and shipped.

Client names are left out. Ask the agent about any project for the details.

Agentic AIEnergy utility · 2026

Multi-agent service automation

An agentic platform that classifies incoming customer requests from email, web forms and tickets, gathers context from CRM, ERP and ticketing, and handles cases on its own or with human approval.

PythonLangGraphPydanticPGVectorPostgreSQL
  • Supervisor, classification, research, processing and quality-check agents
  • PostgreSQL checkpointing, human-in-the-loop approvals, resumable cases
  • Structured extraction from emails and attachments
  • RAG over process documentation and historical cases
  • Introduced AI-assisted development (Claude Code, Copilot) in the team
Document AI · MLPostal & logistics · 2025

Intelligent invoice processing

Extraction, 3-way match against orders and goods receipts, anomaly and duplicate detection, and hand-off of verified invoices to the ERP, with retraining from clerk corrections.

Azure OpenAIDocument IntelligenceLightGBMDatabricksMLflow
Conversational AIIntralogistics · 2025

Fleet management AI assistant

A portal assistant answering fleet questions in plain language with streamed answers, generated charts and deep links, using MCP tools and sandboxed text-to-SQL for exact numbers.

Spring AIA2AMCPdeepevalAngular
RAGInsurance · 2024

AI coverage check for claims

A real-time coverage recommendation when a claim is reported, citing the relevant clause of the policy terms. Uncertain cases go to a claims handler with a confidence score.

Java 21Spring AIGPT-4opgvectorOCR
Full-stackLogistics · 2023–24

Fleet telematics portal

A multi-brand micro-frontend portal for forklift telematics with live sensor dashboards and 3D views, on a Java/Spring Boot backend on Azure.

AngularNgRxModule FederationSpring Boot
ModernisationEnergy utility · 2021–23

Document management modernisation

Moved a JBoss monolith to Spring Boot microservices on Azure, with a new Angular frontend, Camunda workflows and Azure DevOps pipelines.

Spring BootAzureCamundaAngular
Backend · cloudRail transport · 2018–21

Passenger information system

A high-traffic system showing train status, delays and journey chains on many kinds of devices, with Kafka microservices and serverless components on AWS.

JavaSpring BootKafkaAWS Lambda
Plus more than a decade of full-stack, mobile and enterprise projects across automotive, logistics, travel, finance and telecommunications.
About

An engineer first. AI is how I solve problems today.

I hold a doctorate in engineering and have spent more than 15 years building software, from mobile apps and enterprise portals to event-driven backends in Java and Spring Boot.

Since 2024 my focus has been production AI: RAG, multi-agent systems, MCP and classic machine learning, built in Python and Java and integrated into the systems companies already run. I care about the unglamorous parts that make AI trustworthy: grounding, structured outputs, evaluation, and a human in the loop.

I've worked in transport and logistics, insurance, energy, automotive, telecommunications and the public sector. I also train and coach teams.

AvailabilityNow · full-time
Working modelRemote, on-site as needed
LanguagesGerman · English · Arabic
BasedDarmstadt area, Germany
Working together

Three ways to start.

Discover

Architecture & proof of concept

Requirements analysis with business and IT, a solution architecture, and a working PoC on your data.

  • Use-case and data assessment
  • Architecture and risks
  • PoC you can demo
Most common

Build & integrate

Full-time on your team to design, build and ship an AI system into production, integration included.

  • Agents, RAG, ML models
  • ERP / CRM integration
  • Evaluation, monitoring, CI/CD
Enable

Training & coaching

Bring your team up to speed on LLM engineering and AI-assisted development.

  • Hands-on workshops
  • Review guidelines
  • Coding standards for AI tools
FAQ

Quick answers.

Is the chat agent really answering from your CV?

Yes. It retrieves passages from my CV and case studies and answers only from those and from its tools. If something isn't covered, it says it doesn't know instead of guessing.

Why are there no client names?

Projects are described by industry, not by client. I'm happy to share more in a conversation.

Do you work remotely?

Yes, remote is my default, with on-site visits when they help the project.

Java or Python?

Both. Python for agents, ML and data work; Java and Spring AI when the system lives in a JVM landscape.

What happens to what I type into the chat?

It's used to answer you and is handled as described in the privacy policy. Don't paste anything confidential.

Contact

Got an AI problem worth solving?

Ask the agent to pass on a call request. It collects your name, email and topic, and I'll get back to you. Or have a look at my code first.

GitHub ↗