Required skills
- GraphStorm
- Deep Graph Library (DGL)
- AWS SageMaker
- AWS CloudFormation
- Pulumi
- Python
- Graph Neural Networks
- Amazon Neptune
- Gremlin
- MLOps
- PyTorch
- Data Integration
Languages: German · English
Job description
A transport company is seeking a Data Scientist to develop and operate innovative Machine Learning and AI solutions, focusing on fraud detection and process optimization. The role involves building scalable AI/ML models using Graph Neural Networks and Gradient Boosting on AWS.
Requirements
(must be met)
- Production experience with GraphStorm or DGL (Deep Graph Library) including graph construction and data processing
- Production experience with AWS SageMaker, especially pipelines, processing and training jobs, Model Registry, and production endpoints Experience level in years Intermediate (> 3 years experience)
- Professional experience in Data Engineering and/or MLOps Engineering with production-operated data and ML solutions (in 2 projects)
- Practical experience with Infrastructure as Code using AWS CloudFormation or Pulumi
- Proven experience in the development and production operation of automated data pipelines for structured and unstructured data
Nice to have
- Experience with event streaming and event-driven data processing using Apache Kafka or AWS MSK
- Experience with XGBoost including integration into production ML processes
- Experience with Explainable AI methods (SHAP) including integration into production ML processes
- Experience with AI agent systems, Model Context Protocol (MCP), and integration of external systems and observability (Langfuse)
- Experience with distributed data processing frameworks (PySpark, Delta Lake) and data integration via AWS S3 and Azure Blob Storage
Project details
01.01.2027
End: 31.12.2029
Scope: 220 PT
Remote & Frankfurt am Main (30 PT onsite)
Project Language: English
Industry: Transport & Logistics
A company in the transport sector is developing and operating innovative Machine Learning and AI solutions as part of its digitalization efforts. Key application areas include fraud detection in online sales and optimization of operational audit processes. The project's goal is the development, production deployment, and continuous optimization of scalable AI/ML models. Focus areas include Graph Neural Networks, Gradient Boosting methods, automated data processing, and the setup and operation of high-performance MLOps and cloud infrastructures on AWS.
The solutions are intended to ensure reliable, traceable, and efficient production operation while considering existing governance and compliance requirements.
Task Description:
- Development and operation of scalable data pipelines for processing structured and unstructured data using Python, PySpark, Delta Lake, Parquet, and polars/pandas, as well as integration of data sources via AWS S3 and Azure Blob Storage.
- Conception and implementation of data processing processes for Machine Learning, including data preparation, feature engineering, and graph construction from tabular data for Graph Neural Network models (GraphStorm, DGL, PyTorch).
- Setup and further development of graph databases based on Amazon Neptune and Gremlin, including data modeling, bulk loading of large datasets, and optimization of data access.
- Conception, implementation, and optimization of relational database solutions based on Amazon Aurora, Amazon RDS, and PostgreSQL, including schema design, data integration, and query optimization.
- Development and operation of automated data processing and ML pipelines on AWS SageMaker, including processing and training jobs, hyperparameter optimization, and integration of model artifacts into the Model Registry.
- Development and technical integration of production ML components based on Graph Neural Networks and Gradient Boosting (XGBoost), including integration of inference endpoints and implementation of explainability methods (SHAP).
- Development and integration of data and inference interfaces using AWS Lambda, API Gateway, and Step Functions, as well as implementation of event-driven data processing via Kafka/MSK.
- Automation of data-related infrastructure and development processes using Infrastructure as Code (CloudFormation, Pulumi) and implementation and maintenance of CI/CD components in GitLab for data pipelines and ML applications.
- Technical integration of AI-based agent systems, including data connection to external systems (GitLab, ITSM), integration of the Model Context Protocol (MCP), and connection to monitoring and observability solutions using Langfuse.
- Development of data visualizations and analysis dashboards based on Grafana and Plotly for monitoring data processing and presenting relevant analysis and model results.
- Quality assurance, performance optimization, and technical documentation of the developed data architectures, data pipelines, and ML components, considering existing development standards as well as governance and compliance requirements.
- Start: 2027-01
- Duration: 36 Monate
- Remote: hybrid
- Location: Frankfurt, Deutschland, DE
- Languages: German, English
Indexed by 9am's meta-search. You are a job provider? Claim or manage this listing
Found on FreelancerMap DE