Thesis: Hybrid Multi-Agent Orchestration – Evaluating Gas Town & Gas City in Enterprise Environments

Agent-based software development creates opportunities for automation but also raises questions about infrastructure capacity, operating costs, and data protection. This thesis will assess the technical and business feasibility of using Gas Town and Gas City in enterprise environments. A proof of concept combining local computing and private cloud infrastructure will help evaluate how tasks can be assigned to different models to balance cost, performance, and security.

The assignment

• Design and deploy a repeatable proof of concept for an internal software factory using Gas Town/Gas City and the Beads data and memory system.
• Experiment with task routing between CPU-optimized models, small GPU-based models, large local open-weights models, and commercial or private-cloud frontier models.
• Run standardized software development scenarios across local and hybrid hardware configurations, including a 128GB GB10 Grace Blackwell system.
• Measure API and token costs, execution times, compute utilization, hardware throttling, prompt configurability, and code validation results.
• Assess how local and hybrid deployments address enterprise data governance risks compared with cloud-only configurations.
• Produce an enterprise evaluation matrix and an architectural guide linking agent tasks to suitable model sizes and hardware.

Who we are looking for

• Studying towards a bachelor's or master's degree in Computer Science or an equivalent field, with graduation planned for June 2027.
• Fluent in both Swedish and English, in writing and conversation.

Practical details

Based at Knightec Group's office in Sundsvall. Start in January or March 2027. Apply through the company's careers page as soon as possible, and no later than 30 November 2026. Questions can be sent to johanna.edstrom@knightec.se.