Building Large Language Models in the Cyber Domain

Follow Software Engineer Bradley Hartlove’s work on SealingTech’s Operator X, a generative AI platform, as he discusses the fundamentals of building LLM applications.

At SealingTech, we’re always looking for ways to share our knowledge and expertise with our customers and the cybersecurity community. As a Software Engineer at SealingTech, I help develop new innovations that enhance our customer capabilities and their missions in support the cyber warfighter. 

One of those innovations is Operator X™: SealingTech’s new generative AI platform that leverages agents and unifies a cyber operator’s entire toolkit into a single conversational interface. I wanted to take what we learned from building Operator X and teach development teams and fellow engineers how to integrate large language models (LLMs) into their systems.  

If you’re interested in integrating with LLMs but may not have a background in artificial intelligence and machine learning (AI/ML), this video series will help you better understand the core components and common practices used to build LLM applications.  

My hope is that instead of having to sift through the vast sea of libraries and information out there, these short videos will provide a shortcut to a foundational understanding of what LLMs are and how to successfully use them in your applications. 

Let’s get started. 

Introduction to Building LLM Applications 

In this video, we’ll focus on the fundamentals of building large language models including: 

• Learning about the background of LLMs 

• Diving into the process of Tokenization 

• Running through a simple inference 

Inference Engines and Prompting Techniques 

In this video, you’ll learn: 

• How to use the vLLM inference engine 

• Constraining LLM outputs 

• Leveraging few-shot prompting to better guide outputs 

• Prompting a model to think step by step 

Stay tuned for Part II of Bradley’s video series.  

Up next: learn how to build RAG from scratch and LLM Agents. Interested in seeing a demo of Operator X? Contact our team.