13.11.2026 aktualisiert


nicht verfügbar
info: nicht verfügbar bis 31.12.2026
Data Scientist and Software Developer (Python | SQL | R | AWS | Modelling & Analytics)
Chemnitz, Deutschland
Chemnitz +250 km
Dr. rer. nat.Über mich
I help organisations turn complex data into reliable, production-ready solutions – from data platforms and REST APIs to interactive web applications and AI-powered tools. Proven track record in regulated and research driven environments (finance, biotech, public institutions).
Skills
RAGGitOpsLangChainJavaScriptAPIsAmazon Web ServicesBash ShellDatenvisualisierungGitHubR (Programmiersprache)PythonPostgreSQLLinux ServersMachine LearningMySQL
Freelance data scientist and software developer specializing in data platforms, machine learning and generative AI, with a focus on the life and environmental sciences. I help organizations turn complex data into reliable, production-ready solutions – from **data pipelines and REST APIs to interactive web applications, knowledge graphs and AI-powered tools**. Proven track record in regulated and research-driven environments (agriculture and biotech, finance, public institutions), with a strong focus on robustness, maintainability, and clear communication.
Professional Skills
- Analytical thinking
- Independent problem solving
- Creativity
- Effective communication
- Scientific expertise (biodiversity, ecology, plant science)
Programming
- Python, R, SQL, JavaScript, M (Power Query language)
- HTML, CSS, XML
- Package development (Python & R)
- API design (FastAPI, Flask)
- ETL / ELT pipelines, web scraping
- Scripting & workflow automation
- Testing (pytest, testthat)
Generative AI
- Retrieval-augmented generation (RAG, GraphRAG), knowledge graphs
- LLM agents and tool use
- LLM evaluation and observability
- Technologies: LangChain, LangGraph, Langfuse, AWS Bedrock, Google Gemini, vector databases
Machine Learning
- Classification & regression: linear models, tree-based methods, neural networks
- Hyperparameter tuning, cross-validation, model evaluation
- Bayesian modeling
- Dimensionality reduction, clustering
- Technologies: XGBoost, scikit-learn, caret, tidymodels, PyTorch, NumPy, dplyr, pandas, JAGS, Stan
Databases
- Data modeling, schema design, data normalization
- Query design and optimization
- OLTP & OLAP systems
- Technologies: PostgreSQL, Supabase, MySQL, SQLite, DuckDB, Oracle DB, Neo4j, Databricks
Geospatial
- Spatial data processing and analysis
- Mobile field data collection
- Technologies: QGIS, QField, PyQGIS, GDAL, GeoPackage, PostGIS, sf, terra
Visualization & Reporting
- Interactive reports
- Dashboards, graph visualization
- Automated documentation
- Technologies: R Shiny, Plotly Dash, Databricks dashboards, Quarto, Markdown, Plotly, ggplot2
Development & Orchestration
- Version control: Git, GitHub, GitLab, CI/CD
- Cloud: AWS, GCP, Databricks
- Containers & environments: Docker, pip, conda, venv, renv, uv
- Workflow automation: make, targets, Databricks Jobs, Camunda
- OS & scripting: Linux, Bash, Windows
- Agile tools: Jira, Confluence, Slack, Teams
Erfahrung
Berufserfahrung5 - 10 Jahre
Freelancer seit2022
Sprachen
DeutschMutterspracheEnglischverhandlungssicherFranzösischGrundkenntnisse
Projekthistorie
Syngenta's Product Biology team is working on a GraphRAG system for crops, crop pests, regulation and
phytosanitary standards. The work is foundational for a broader plan to integrate large volumes of data and
documents at Syngenta into a structured knowledge base for decision-makers, researchers and agentic
systems. As an external consultant in this ongoing effort, I am auditing the current setup and contributing to its
development in areas such as data modeling, data retrieval and agent evaluation.
For the state forestry agency of Brandenburg, I build data quality tooling and field data collection systems for two
forest monitoring programs. For the intensive forest environmental monitoring (ICP Forests Level II), I developed
a multi-user Shiny application in which domain experts review and correct soil sensor time series (matric
potential, soil water content) against hydrological model simulations. Every review decision is stored in a
dedicated PostgreSQL review database. For the forest regeneration survey (Verjüngungsmonitoring), I
developed a code-generated QGIS/QField project for structured, offline-capable data capture on tablets in the
field. Both projects make quality assessment and field collection systematic and traceable, and reduce manual
rework.
As part of an ongoing engagement with KWS Saat, I have been building and extending a Python-based discovery platform for plant breeding research. The work spans backend API engineering, interactive analytics application development, and AI-assisted research workflows, with responsibilities ranging from schema design and validation to CI/CD and deployment hardening. The resulting platform helps researchers explore complex biological relationships and access advanced analysis tools through a practical, user-facing interface.