Passionate about the intersection of Artificial Intelligence and Geospatial systems — where machine perception meets the physical world. I build end-to-end platforms that bridge satellite and drone imagery, cloud-native geospatial infrastructure, and large language models into cohesive, operational intelligence pipelines.
SPECTRA is the result of that obsession: a homelab-scale system that runs production-grade geospatial AI — from raw drone COGs through a STAC catalog, object detection, LLM scene enrichment, vector search, and a natural-language map interface — on a three-node K3s cluster sitting on my desk.
Geospatial Intelligence
Cloud-native geospatial stacks — STAC, COG, TiTiler, OGC APIs — and making Earth observation data accessible at scale.
Applied AI
Vision models, LLM enrichment pipelines, and agentic systems that reason over real-world sensor data rather than synthetic benchmarks.
Hybrid Search & RAG
Combining semantic vector search with structured metadata filtering to build retrieval systems that understand both meaning and context.
Homelab Infrastructure
Running real production workloads on air-gapped K3s — Cilium eBPF networking, GitOps, Sealed Secrets, and zero-trust cluster design.
Tools and technologies powering SPECTRA and the homelab behind it.