Data platforms & pipelines
Streaming and batch pipelines, warehouse and lakehouse design, and microservice architectures on Kubernetes and the cloud. Built for real-time workloads and for teams that need to onboard new sources fast.
Data & AI engineering · Munich, Germany
Data and AI platforms your team can run without me.
I'm Feras, a lead data engineer with more than ten years across energy, manufacturing, IoT and telecoms. I design the platforms, pipelines and machine-learning systems behind real-time decisions, from the factory floor to Europe's power grid.
- Currently Lead Data Engineer at E.ON Digital Technology
- Remote across Europe, on site in Munich and Berlin
Experience at
- E.ON
- FactoryPal
- Nokia
- Telekom Innovation Labs
- MCS Data Labs
Services
Three ways I help teams ship data and AI work.
Whether you need a platform built from scratch, a model taken out of a notebook, or a senior second opinion before a big decision.
AI & machine-learning systems
Forecasting and anomaly detection on time-series data, and LLM applications with retrieval over your own documents. I take models out of notebooks and into versioned, monitored production services.
Advisory & team enablement
Architecture reviews, platform roadmaps, build-versus-buy decisions and hands-on coaching for your engineers. For leaders who need a senior voice in the room without a full-time hire.
Project delivery
Fixed scope, fixed price. A working system with docs and handover.
Fractional lead
One to three days a week embedded with your team for a few months.
Advisory day
A focused day to review an architecture, a plan or a hiring decision.
Selected work
Systems I have designed and built.
A few representative projects. Details are kept general out of respect for the companies involved. I'm happy to go deeper on a call.
Real-time congestion detection for the electrical grid
Designed and built the data platform that detects congestion on E.ON's grid in real time, and the microservice architecture that lets additional grid subsidiaries onboard quickly. Contributed to the architecture of the group's private cloud platform with a focus on scalability and security.
Time-series platform and MLOps for shop-floor analytics
Built scalable systems for multi-dimensional time-series data on AWS and Kubernetes, near-real-time processing with Apache Flink, and the DataOps and MLOps platform the data-science team used to ship models: MLflow, Airflow, Evidently and TensorFlow Serving.
Real-time analytics for wearable sensor devices
End-to-end system for streaming data from wearables: MQTT ingestion, NiFi data flows, InfluxDB storage and Grafana alerting, deployed on containerised distributed infrastructure. Also wrote firmware drivers for the optical, motion and gas sensors on the device.
Forecasting and anomaly detection research
Developed and evaluated deep-learning and statistical models for time-series forecasting and anomaly detection (LSTM, autoencoders, SARIMA), and built a component of a data-driven network intrusion detection system based on traffic analysis.
How I work
No surprises, working software early, and a clean handover.
-
1
Intro call
Thirty minutes, free. We talk through the problem and decide together whether I'm the right person for it.
-
2
Scoping
A short written proposal: goals, an architecture sketch, timeline and price. You know what you're buying before you commit.
-
3
Build in the open
Weekly demos, working software from the first weeks, and decisions written down so nothing lives only in my head.
-
4
Handover
Your team owns the result: documentation, runbooks, monitoring and pairing sessions until they're comfortable running it.
Principles
- Boring technology first. Proven tools unless there's a clear reason not to.
- Observability from day one. If you can't see it, you can't run it.
- Your team should not need me. Success is when you can run and extend the system without me.
About
I learned this stack from the bottom up.
I started by building the Palestinian Museum's computer network from nothing, switches, firewalls, servers and all. Then I moved to Budapest for a master's in data science, did machine-learning research at Telekom Innovation Labs and Nokia, and spent the last six years building data and ML systems in Berlin and Munich.
That path is why I care about the whole stack: the model, the pipeline that feeds it, the cluster it runs on, and the person on call when it breaks at 3 am. I like working with teams who want things to be simple, well understood and genuinely finished.
Outside of work I'm usually learning something new, most recently retrieval-augmented generation and agentic coding tools, or somewhere in the Alps.
He led the development of the Palestinian Museum's computer network from scratch. This entailed a plethora of challenging activities, from designing and implementing a sophisticated local network to putting in place the required servers and securing the entire network. He spared no effort or time to make this infrastructural establishment see the light, and succeeded in creating a common ground and a common language with professionals from different walks of life.
Questions
Things people usually ask first.
Do you work remotely or on site?
Mostly remotely, across European time zones. I'm based in Munich and regularly in Berlin, so on-site kick-offs and workshops are easy to arrange.
How do you price your work?
Projects are quoted as a fixed price for a fixed scope. Fractional and advisory work is billed by the day. You get a written quote after the intro call, never a surprise invoice.
Can you work with our existing engineers?
That's my preferred setup. I pair with your team, review their work, and leave documentation and runbooks behind so the system stays healthy after the engagement ends.
Which stacks are you comfortable with?
Python-first, with Kafka, Flink, Spark, Airflow, Kubernetes, AWS and Azure. On the ML side PyTorch, TensorFlow and MLflow, plus LLM tooling for retrieval-based applications. If you use something else, ask. I've switched stacks before.
What size of company do you work with?
Startups that need senior hands fast, mid-sized companies modernising their data stack, and enterprise teams that need a specialist for a defined piece of work.
Let's talk
Have a data or AI problem you want solved properly?
Book a free thirty-minute call. Bring the messy version of the problem. I'll tell you honestly whether I can help and what it would take.