Deniz Akdeniz

Subterra AI specializes in using photogrammetry and advanced imaging technologies to build 3D models of Subground. The platform focuses on sewer and manhole inspections, leveraging AI and human interaction to provide actionable insights and streamlined workflows.

In my role, I contribute to nearly every aspect of the product lifecycle, including:

• Backend Development: Creating and maintaining AWS Lambda functions in Python and Node.js.
• Frontend Development: Building and enhancing user interfaces with React and TypeScript.
• AWS Management: Overseeing serverless architecture and resource configurations.

The platform processes standard and 360° equirectangular videos from sewer and manhole inspections, converting them into structured file formats, professional PDF reports, and GIS-integrated web presentations.

Key Achievements:

• ArcGIS Integration: Developed integrations with ESRI ArcGIS Online and ArcGIS Enterprise, enabling clients to display inspection-related layers and export results directly to ArcGIS layers, significantly improving data management and accessibility.
• Video Merge and Slicing Automation: Designed tools to merge and slice GoPro videos for sewer inspections. Previously, users manually merged videos, matched GIS data, and sliced recordings for each pipe segment locally. The new functionality allows multi-video uploads, automatic merging with manual ordering options, GIS-based matching, and direct initiation of multi-segment inspections—all processed on the platform. This was implemented using AWS Lambda functions in Node.js, triggered by CloudFormation scripts, streamlining video preparation and inspection workflows.

- Created backend services for GIS (location) based web application with Django and Django Ninja Rest Framework and PostgreSQL DB models. - Developed integration between the application and Quickbase providing bidirectional data flow and synchronization. - Implemented Auth0 authentication for API calls. Developed remote API service integrations for spatial data ingestion with optimized costs. - Used Jupyter Notebooks to create/convert/test tabular and geospatial data for the backend application. - Developed simple React UI applications to test and experience the data/workflow between the backend and frontend. - Created/Revised Docker and K8s manifest files for AWS deployments.