Kirkland & Ellis Plans to Build Its Own Legal AI System With Fine-Tuned Open Source Models
Kirkland & Ellis is laying groundwork to develop an internal legal AI platform rather than relying on commercial off-the-shelf tools, according to job postings that reveal the firm's technical approach to its $500 million innovation project.
Two new AI Infrastructure Director roles posted May 27 demand experience managing "on-premise GPU environments"-the specialized hardware needed to fine-tune open source language models with proprietary firm data. The positions, based in Houston and Chicago, offer salaries between $302,000 and $335,000.
What the Job Postings Reveal
The infrastructure director role requires candidates to "design, manage, and optimize the firm's AI infrastructure - spanning on-premise GPU environments and Microsoft Azure-based AI platforms." The emphasis on owning GPU clusters suggests Kirkland intends to train models using its own data rather than relying on third-party services.
The job description mentions "ML services" and "governed environments for experimentation"-language consistent with building and testing custom AI systems. The firm also seeks "AI Innovation Advisers" across multiple U.S. locations at salaries ranging from $153,000 to $220,000.
These advisers will "embed within practice groups to translate legal tasks and workflows into scoped AI solutions" and work directly with engineers on building and testing AI-driven processes. Job postings specify hands-on experience with platforms like Harvey, Legora, CoCounsel, and Lexis+ AI.
Scale of the Effort
Kirkland's jobs site lists approximately 85 open positions mentioning AI, with postings dating back to March and new roles added regularly. The firm said it would deploy around 180 people on the project, though the volume of active postings suggests continued hiring.
The breadth of roles-from infrastructure engineers to legal workflow specialists-indicates the firm is serious about this undertaking. Building a custom legal AI system requires both technical expertise and deep understanding of how lawyers actually work.
Why This Matters
Most law firms use commercial legal AI tools built by Thomson Reuters, LexisNexis, and others. Kirkland's approach differs: the firm wants to move beyond what it calls the legal AI "floor"-standard platforms with limited customization.
Fine-tuning an open source model with proprietary firm data and on-premise infrastructure could offer advantages beyond performance. A system running on Kirkland's own servers may provide greater data privacy than cloud-based alternatives, even though enterprise solutions typically include security controls.
Whether clients will perceive meaningful differences between Kirkland's custom system and competitors using third-party tools remains unclear. Competitors like Latham & Watkins and Skadden have embedded commercial platforms alongside custom point solutions.
The project's success will depend on whether the additional investment yields measurably better legal work or primarily serves as a data privacy differentiator. The finished product hasn't yet been tested against existing alternatives.
What This Signals
Few firms have the capital or technical depth to replicate Kirkland's approach. The $500 million commitment and hiring scale suggest the market for custom legal AI infrastructure remains limited to the largest practices.
Kirkland's public recruitment effort does underscore how seriously major law firms now view AI as central to competitive positioning. Whether building custom systems or optimizing commercial ones, legal professionals increasingly need to understand how these tools work and where they succeed or fail.
For legal professionals interested in AI implementation, understanding AI for Legal and the fundamentals of Generative AI and LLM provides context for evaluating these emerging systems.
Your membership also unlocks:
