Hello, I am Bakar, a geospatial professional with a passion for all things spatial.

Professional experience

1. 11/2025 – present

   Geospatial Data Specialist Beagle Systems, Hamburg

   Beagle Systems flies long-range drones beyond visual line of sight (BVLOS) to monitor pipelines, utilities and other infrastructure.

   • Developed an open-source drone LiDAR pipeline that georeferences point clouds with PPK, classifies power-line scenes, rebuilds missing conductors from their sag and rates nearby vegetation by risk; tree-fall analysis in progress.
   • Built an AI-driven right-of-way and construction-site monitoring pipeline: deep-feature-matching georeferencing, a fine-tuned foundation model for detection, an open-source VLM to verify it, and real-time inference on an NVIDIA Jetson.
   • Fine-tuned a building-segmentation foundation model on in-house labels created from 10 cm orthophotos and scaled it across the Cologne area.
   • Built a PostGIS and TiTiler backend that serves large GeoTIFFs to the frontend as map tiles, and supported flight planning with pre-flight checks and new geodatabase data.

2. 03/2024 – 10/2025

   Geospatial (EO) Data Scientist, part-time Karuna Technology UG, Germany

   • Fine-tuned a pretrained deep-learning model on in-house training data to map urban tree canopies from aerial imagery, then tracked their health with Planet NDVI and NDWI time series in Google Earth Engine to flag underperforming trees.
   • Projected the long-term ecosystem services of the mapped trees through a tree-benefits API, from carbon storage to further environmental benefits, estimating the urban canopy’s value over the coming decades.
   • Built Google Earth Engine workflows that clean multi-year Sentinel-1 radar time series with speckle filtering and apply an equation to derive soil-moisture maps, showing where golf courses suffer from water stress.

3. 04/2021 – 02/2024

   GIS Specialist & Data Engineer MapTransfer, Germany

   • Managed the end-to-end geodata workflow for a construction company: georeferenced construction plans and orthophotos, converted them into cloud-optimized GeoTIFFs, uploaded them to AWS S3, and set up QGIS with PostgreSQL as the customer’s secure front end.
   • Designed and authored several GIS courses for students and professionals, covering LiDAR processing, spatial analysis, Spatial SQL, ALKIS and GeoServer.
   • Converted AutoCAD data in QGIS into accurate, ALKIS-compliant GIS layers for telecom infrastructure projects.
   • Built interactive Web GIS applications in OpenLayers, serving WMS layers through GeoServer from PostgreSQL.

Skills

Geospatial AI

Fine-tune foundation models for detection and segmentation on drone and aerial imagery, and run open-source LLMs and VLMs as quality gates in production pipelines, from GPU servers down to the edge.

LiDAR

Take drone LiDAR from raw scans to deliverables: PPK georeferencing, sensor calibration, point-cloud classification and power-line analytics.

Remote sensing

Turn SAR, multispectral and hyperspectral data into maps that answer questions, from Google Earth Engine time series to centimeter-scale drone imagery.

Spatial data

Model, clean and convert vector data at scale, from ALKIS cadastre and CAD plans to analysis-ready GIS layers, in GeoPandas, PostGIS and QGIS.

Data engineering

Build the plumbing behind it: PostGIS databases, cloud storage on AWS, containerized pipelines and tile services that feed web frontends.

Programming

Python every day; Spatial SQL, R and Bash where they fit.

Geology

Trained geologist with a BSc and MSc: numerical modeling and petrophysics.

Languages

English

C2

German

B2