Home / Companies / Crowdstrike / Blog / Post Details
Content Deep Dive

How CrowdStrike Trains GenAI Models at Scale Using Distributed Computing

Blog post from Crowdstrike

Post Details
Company
Date Published
Author
Andrei Preda - Alexandru Dinu - Florian Stortz - Nathan Nusaputra - Catalin-Andrei Stan
Word Count
2,855
Company Posts That Month
16
Language
English
Hacker News Points
-
Post removed?
No
Summary

CrowdStrike is significantly advancing the training of large language models (LLMs) for cybersecurity applications by leveraging distributed computing and cloud-based infrastructures. As threats evolve with the integration of LLMs in cyber attacks, CrowdStrike has made it a strategic priority to develop custom LLMs tailored for cybersecurity challenges. Utilizing resources such as the Google Cloud Vertex Training Platform, the company efficiently manages the training of these models at scale, employing techniques like data, tensor, and pipeline parallelism to optimize resource use and performance. The company focuses on addressing practical challenges in LLM training, such as data diversity and memory management, by implementing synthetic data augmentation and gradient checkpointing. These efforts are part of a broader initiative to enhance the capabilities of their cybersecurity solutions, ensuring they remain at the forefront of AI-driven threat detection and response. CrowdStrike's ongoing research and infrastructure investments aim to improve the efficiency and scalability of their machine learning models, ultimately strengthening their ability to preemptively counteract sophisticated cyber threats.

Trends Found in this Post
Trend Post Mentions Total Month Mentions Posts Companies MoM
LLM 24 4,308 744 242 -15%
AI Agents 2 3,387 723 216 -28%
Zero Trust 2 202 54 28 +66%
AI Coding Assistant 1 721 236 105 -30%
AI Guardrails 1 430 152 53 -24%
AI Model Fine-tuning 1 684 149 78 +46%
Data Pipeline 1 1,051 283 78 +133%
Observability 1 2,935 607 185 -3%
Use This Data

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.