Novikov on local LLM infrastructure and security
Evgeny Novikov delivered a technical talk on local LLM deployment infrastructure and security practices at Xecut Hackerspace, with slides released in markdown and HTML formats.
Evgeny Novikov presented a technical session on local large language model infrastructure and security at Xecut Hackerspace on June 19, 2026, addressing the operational challenges practitioners face when running uncensored models on their own hardware. The talk, recorded and published on YouTube, covers deployment architecture, security considerations, and the infrastructure layer that sits between raw model weights and production workloads. Slides are available in both markdown source and browser-ready HTML formats, making the material accessible to developers who prefer different documentation workflows.
Local deployment eliminates the server-side safety filters and usage policies that govern commercial API providers, but introduces new operational burdens: hardware provisioning, model quantization choices, inference optimization, and network isolation for sensitive workloads. Novikov's talk addresses these practitioner concerns directly, filling a gap in public documentation that often focuses on model capabilities rather than the infrastructure required to run them reliably. The markdown source allows contributors to fork and adapt the content, while the HTML build provides immediate reference for teams evaluating local deployment strategies.
