Portfolio | Silas Pignotti

Geo Data Scientist ยท AI Engineer

M.Sc. Geoinformation student at BHT Berlin. Building geospatial ML systems and AI tooling.

What I Do

  • Geospatial Data Science

    Machine learning and deep learning on satellite imagery and geodata. End-to-end pipelines from raw data to trained models, with a focus on reproducibility and evaluation.

  • AI & Automation

    Context Engineering, agentic workflows, and LLM-based tooling. I build structured prompt systems, automated pipelines, and integrate AI into development and analysis workflows.

  • Open Source

    Python CLI tools and web applications on PyPI and GitHub. Built to solve real problems in my own work, released for others to use.

Skills & Technologies

Core tools and platforms used across geospatial analysis, machine learning, and production workflows.

Python
SQL
PyTorch
scikit-learn
Google Earth Engine
PostGIS
QGIS
Python
SQL
PyTorch
scikit-learn
Google Earth Engine
PostGIS
QGIS
Python
SQL
PyTorch
scikit-learn
Google Earth Engine
PostGIS
QGIS
LangChain
LangGraph
LiteLLM
Claude Code
OpenCode
n8n
Notion
Codex
LangChain
LangGraph
LiteLLM
Claude Code
OpenCode
n8n
Notion
Codex
LangChain
LangGraph
LiteLLM
Claude Code
OpenCode
n8n
Notion
Codex
GCP
PostgreSQL
FastAPI
Docker
Git
Linux
Prefect
MLflow
GCP
PostgreSQL
FastAPI
Docker
Git
Linux
Prefect
MLflow
GCP
PostgreSQL
FastAPI
Docker
Git
Linux
Prefect
MLflow

Selected projects that represent core strengths across geospatial ML and AI tooling.

View all
  • litresearch project cover image
    AI/Automation 2026 Completed

    litresearch

    Automated literature search from a research question to a curated, deduplicated source set.

    Open-source CLI for automated literature research. From research question to curated paper set with report and BibTeX export. Published on PyPI.

    • Python
    • Typer
    • LiteLLM
    • Semantic Scholar
  • Urban Tree Transfer project cover image
    Geospatial 2026 Completed

    Urban Tree Transfer

    Cross-city transfer of urban tree genus classification. Can a model trained in Berlin classify trees in Leipzig?

    Cross-city transfer of tree species classification using Sentinel-2. End-to-end ML pipeline with spatial block CV across Berlin and Leipzig.

    • Python
    • XGBoost
    • scikit-learn
    • GeoPandas
  • context-engineering project cover image
    AI/Automation 2025 Ongoing

    context-engineering

    Structured context layer for AI-assisted development. Reusable commands, controlled prompts, persistent session memory.

    Reproducible framework for AI-assisted coding with OpenCode. Custom commands, skills, prompt pipelines, and project templates.

    • Python
    • OpenCode
    • Git

Get in Touch

M.Sc. Geoinformation student at BHT Berlin. Open for freelance projects, collaborations, and roles in Geo Data Science or AI Engineering.

Based in Berlin.