I build GeoAI agent systems, spatial ML infrastructure, and deterministic evaluation harnesses: standards-driven Agent Skills for geospatial reasoning, MCP tooling that automates ArcGIS Pro without handing an LLM the runtime, reproducible Sentinel-2 pipelines, and decision systems people actually open.
The through-line is evidence. Every number below comes from a benchmark, a manifest, or a released artifact in the repo it describes, and where a result does not exist yet the repo says so instead of implying one.
| System | What it is | Status | Links |
|---|---|---|---|
| geoai-skills | 18 Agent Skills covering the geospatial lifecycle, routed by an orchestrator | v0.4.0 · MIT · routing benchmarked 2026-08-05 | Repo · Benchmark |
| arcgis-mcp-bridge | MCP server exposing 100 ArcGIS Pro / ArcPy geoprocessing tools to LLM hosts | Apache-2.0 · PyPI · Glama A-rated | Repo · PyPI · Glama |
| benchfck | Deterministic evaluation generator and exact Rust harness for machine-state tasks | v0.4.0-alpha engineering candidate · no model results published | Repo · Validity contract |
| sentinel-crop-pipeline | Reproducible Sentinel-2 preparation: discover, mask, patch, label | v0.3.2 · Apache-2.0 · PyPI · Zenodo DOI | Repo · PyPI · DOI |
| agri-dss | Zero-backend spatial decision support across 5 districts and 147 neighborhoods in Western Antalya | Live | tarimsalkoridor.online · Repo |
| FOUNDER.EXE | Startup simulation encoding Turkish and US tax, grant, SAFE and cap-table rules | Published · Windows x64 · paid | itch.io |
Research repos: kutri-resilience-index, an urban resilience composite indicator with a DOI · turkiye-housing-prices-pandemic, inflation-adjusted HPI and spatial clustering · agri-unet, the downstream U-Net training track for sentinel-crop-pipeline.
geoai-skills ships a frozen routing benchmark instead of a claim about how well the skills "work". 167 cases, 18 skills, run 2026-08-05 on Claude Code 2.1.214 with claude-sonnet-5, against a paired control with the skills switched off.
| Arm | Precision | Recall | Full-route accuracy |
|---|---|---|---|
| Skills enabled | 99.17% | 96.77% | 96.41% |
| Skills disabled (control) | n/a | 0% | 0% |
The control recorded zero activations across all 167 cases, which is what makes the treatment arm mean anything. The benchmark measures routing, not answer quality, and the repo states that boundary explicitly.
arcgis-mcp-bridge has to give an agent real geoprocessing power while keeping a licensed GIS runtime out of the host process. Two isolated processes, two independent path validations, one refusal path.
flowchart TB
H["LLM host / AI agent"] -->|MCP over stdio| S["Async MCP server"]
S --> A["PathGuard A<br/>pre-check + confirmation gate"]
A -->|validated request| W["Isolated ArcPy worker"]
W --> B["PathGuard B<br/>independent re-validation"]
B -->|validated path| G["ArcGIS Pro / ArcPy runtime<br/>100 geoprocessing tools"]
G --> O["Structured NDJSON result"]
A -.->|blocked| X["Rejected safely"]
B -.->|blocked| X
style H fill:#ECF2EC,stroke:#1A221C,color:#1A221C
style S fill:#D7E9DD,stroke:#0B7F47,color:#1A221C
style A fill:#F6E6C8,stroke:#8A5A11,color:#1A221C
style W fill:#DCE5EE,stroke:#2C67A0,color:#1A221C
style B fill:#F6E6C8,stroke:#8A5A11,color:#1A221C
style G fill:#D7E9DD,stroke:#0B7F47,color:#1A221C
style O fill:#ECF2EC,stroke:#1A221C,color:#1A221C
style X fill:#F3D9D4,stroke:#A83A2B,color:#7A2418
- Deterministic evaluation. Exact generators, restricted verifiers, preregistered release gates, and a hard line between engineering evidence and model evidence.
- GeoAI agent systems. Turning tacit spatial practice into portable agent instructions, then measuring whether the routing actually holds up.
- Remote sensing ML. Spatially blocked splits, patch-aligned ground truth, and segmentation experiments downstream of a pipeline that is reproducible first.
Stack: Python · Rust · PyTorch · GeoPandas · Rasterio · ArcPy · PySAL · PostGIS · FastAPI · MCP · Agent Skills · Docker
Open to work on GeoAI agent systems, spatial machine learning, remote-sensing pipelines, MCP infrastructure, and evaluation design.
Portfolio (Türkçe) · LinkedIn · nsduraan@gmail.com
