blackshark.ai

• Build and optimize ML orchestration pipelines that coordinate model training and inference across distributed worker pools • Design resilient, high-throughput services that process terabytes of geospatial imagery through GPU-accelerated inference • Develop the APIs and abstractions that allow customers to chain, filter, and compose AI models for complex detection workflows • Collaborate with ML Researchers to put new models in production • Tackle memory optimization, GPU autoscaling, and resource scheduling challenges unique to large-scale imagery processing

• Build and evolve our frontend application that serves as the primary interface for AI model training, inference execution, and results exploration. • Develop rich, interactive map-based visualizations for displaying detection results, segmentation masks, and training data coverage across large geographic areas. • Design intuitive workflows for complex geospatial operations: imagery upload and management, model configuration, inference parameterization, and results export. • Build performant UI components for browsing and interacting with large-scale imagery handling pan, zoom, and overlay rendering for high-resolution satellite and aerial data. • Work closely with backend engineers to integrate with our FastAPI services, designing responsive interfaces that communicate pipeline status, processing progress, and async job results in real time. • Contribute to our public API developer experience, building documentation portals and interactive tooling for external integrators. • Ensure the application works reliably across a variety of deployment contexts. We run our platform as SaaS, single-tenant cloud, on-premise, and on edge laptops in the field.

• Lead technical discovery with simulation OEMs and defense customers, turning mission requirements and system constraints into clear, executable solution paths • Deliver presentations, demos, and technical workshops that prove capability under operational scrutiny • Design and execute proof-of-concepts that validate performance and integration into customer ecosystems (e.g., Unreal, CDB pipelines, OEM toolchains) • Shape technical responses to RFIs/RFPs and conduct trade studies that position Blackshark as the credible, differentiated choice • Surface technical risks early - performance, interoperability, security, deployment complexity - and resolve them before they become blockers • Work closely with Sales leadership, Product, Engineering, and Delivery to ensure what we propose is technically sound, competitively strong, and ready to ship

• Build and optimize preprocessing pipelines that ingest, tile, and transform geospatial imagery (GeoTIFF, multispectral, SAR, COG) for downstream ML inference • Develop postprocessing services that convert model outputs into production-ready deliverables: segmentation masks, probability maps, and vector detections in GeoPackage format • Design resilient, memory-efficient services for processing large-scale imagery through distributed worker pools • Work on our unified Huntr-to-Replika pipeline, automating the flow from detection outputs to 3D-ready terrain tiles with auto-generated configuration • Tackle challenges around coordinate reference systems, spatial indexing, and data format interoperability • Optimize for the unique characteristics of geospatial workloads: memory-bound processing, batch-heavy operations, and streaming large raster datasets

• Build and operate Kubernetes clusters across AWS, Azure, GCP and bare-metal K3S deployments • Own infrastructure as code using Terraform, evolving our multi-environment architecture • Design and improve CI/CD pipelines to accelerate the path from code to production • Enhance observability across the stack with Prometheus, Grafana, and the ELK stack • Maintain security standards required for defence and government customers • Optimize cloud costs while ensuring performance and reliability • Travel to customer sites to support on-premises deployments and troubleshoot production issues • Ensure testing and staging environments are reproducible and consistent with production