# Junior MLOps Engineer

Location GermanyDesired Salary UnspecifiedWork preference Full Time, ContractLinks Joined19 Jun 2026Field / Industry Software Engineering

* SharePreferences:Status: Actively lookingRelocation: NoNotice Period: ImmediateSkill Assessments:This user has not passed any tests yet

Languages:  Arabic -   English -   German -

Report
[Overview](#overview)

## About Me

I am a junior MLOps engineer focused on ML deployment, pipeline automation, and production-ready machine learning systems. I have hands-on experience building end-to-end ML pipelines in industrial environments and enjoy turning research and prototypes into reliable software.

At Mars GmbH, I designed and shipped a computer vision quality-inspection system that covered the full lifecycle from data ingestion and feature engineering to model training, validation, and integration into live quality-control workflows. This experience strengthened my ability to work across ML engineering, software delivery, and operational constraints.

I work extensively with Docker, FastAPI, Azure ML, MLflow, DVC, GitHub Actions, and Kubernetes. I also build reproducible workflows, modular codebases, and automated retraining and monitoring systems to support maintainable production ML.

In my current research role at TU Dortmund University, I design modular ETL pipelines for high-frequency wearable sensor data and build reusable Python libraries that standardize experiment workflows. I place strong emphasis on reproducibility, structured logging, schema validation, and pipeline reliability.

I have also completed internships in data science and ML engineering, where I delivered regression and classification pipelines, improved model performance through feature analysis and hyperparameter tuning, and contributed to production-quality ML software. These roles helped me develop a practical, end-to-end understanding of the ML lifecycle.

Outside of work, I am building an open-source MLOps portfolio that includes containerized model serving, automated retraining, and production monitoring. I am motivated by solving real-world ML infrastructure problems and creating systems that are robust, scalable, and easy to maintain.

## Skills

### [Python](https://jobicy.com/talent/python.md)[SQL](https://jobicy.com/talent/sql.md)[Kubernetes](https://jobicy.com/talent/kubernetes.md)[Docker](https://jobicy.com/talent/docker.md)[Azure](https://jobicy.com/talent/azure.md)[CI CD](https://jobicy.com/talent/ci-cd.md)[Git](https://jobicy.com/talent/git.md)[PostgreSQL](https://jobicy.com/talent/postgresql.md)[TensorFlow](https://jobicy.com/talent/tensorflow.md)

## Education

Anhalt University of Applied Sciences  Oct 2021 - Jun 2024  M.Sc. Biomedical Engineering

Grade: 1.5. Thesis on feature engineering, machine learning, and computer vision-based approach for pet food chunk quality optimization.

Minia University  Oct 2013 - Jul 2018  B.Sc. Biomedical Engineering

Thesis on Artificial Pancreas Prototype and Automated Insulin Delivery System, with embedded control systems and biomedical signal processing.

## Experience

Research Associate @ TU Dortmund University  Apr 2025 - Present Design and maintain modular ETL pipelines for high-frequency wearable sensor streams; build reusable Python libraries; enforce reproducibility through version control, automated testing, and structured configuration management; apply data engineering best practices such as schema validation, pipeline idempotency, and structured logging.

ML Engineering Intern @ Mars GmbH  May 2023 - Apr 2024 Built and deployed a production computer vision pipeline using Detectron2 to automate pet food quality inspection; engineered 85+ geometric feature descriptors; designed the full ML lifecycle from preprocessing to live QC integration; reduced manual QC time by about 40%; managed Azure ML compute infrastructure; collaborated with engineering and quality teams.

Data Science Intern @ ZUMMIT INFOLABS  Oct 2022 - Jan 2023 Delivered end-to-end regression and classification pipelines; improved baseline model performance by about 10% through systematic feature analysis and hyperparameter tuning; produced reproducible client-facing outputs.