Everlaw Is Hiring 33 People to Build the AI Stack That Stops Lawyers From Getting Sanctioned

What Deep Dive Actually Does

In October 2025, at its annual Everlaw Summit in San Francisco, CEO and founder AJ Shankar took the stage to announce the general availability of EverlawAI Deep Dive, a generative AI feature that lets legal teams ask natural-language questions across entire document collections and get back citation-backed answers in seconds. The tool spent eight months in beta, tested against matters containing tens of millions of documents. It reaches full customer availability before the year ends.

The pitch is straightforward: instead of keyword searches that return thousands of documents to sift through, a lawyer can ask "What did employees say about expense reporting?" and receive a ranked answer with direct citations to the source material. Deep Dive runs on retrieval-augmented generation (RAG), pulling relevant passages from the document corpus before the large language model generates a response. Everlaw pairs this with LLM reasoning models that break complex questions into smaller logical steps. The system stays constrained to the user's own data. If no document in the corpus supports an answer, it says so. It does not guess.

That constraint is the core design choice. Shankar drew the distinction explicitly during his keynote: "Ask an LLM 'why the sky is blue,' and it will use its embedded knowledge to answer. That's not helpful when you're trying to make an argument supported by hard evidence from within your discovery universe. Worse, if the LLM doesn't know the answer, it may make something up." Deep Dive, by contrast, returns a message that it cannot find evidence in the corpus. Every answer it does produce is ranked by confidence level and accompanied by facts and referenceable source documents.

The beta program ran for roughly eight months, starting with a closed phase and expanding to open beta in August. During that period, beta customers used Deep Dive on projects exceeding 1 million documents. The largest enrolled matter surpassed 10 million. The average database size across beta matters was approximately 166,000 documents. More than 250 customers now use some part of Everlaw's generative AI suite, including federal agencies and participants in the company's nonprofit program.

The practical use cases cluster around speed. During early case assessment, a team receiving a new production from opposing counsel can interrogate the full corpus on day one rather than waiting for first-level review to surface key facts. For deposition and trial prep, attorneys can pull specific quotes and facts tied to particular witnesses or contract terms without manual review. Julie Brown, director of practice management at Am Law 200 firm Vorys, described using Deep Dive on a 2 million-document collection with a week-long production deadline as a quality control pass; the tool identified relevant documents the team's other review methods had missed.

Deep Dive is not a standalone product. Shankar emphasized that it works as one component within the broader Everlaw platform, alongside Coding Suggestions, Clustering, and Storybuilder. Pricing follows a one-time per-gigabyte ingestion fee that unlocks unlimited questions for the lifetime of a case. Three other AI features (Review Assistant, Writing Assistant, and Deposition Analyzer) folded into Everlaw's core per-gigabyte rate at no additional charge as part of the same announcement, and the company cut the price of batch Coding Suggestions by more than 40%.

The signal underneath the product launch: Everlaw is betting legal AI has crossed from experimental to operational. The pricing restructuring removes per-use metering on several features, an admission that credit-based friction was slowing adoption. Shankar framed it directly: "We know how hard it is for you to operationalize the use of these really powerful tools with a system where every usage is metered." The engineering effort to get there, and the roles Everlaw is hiring to sustain it, tell the rest of the story.

Inside Everlaw's 33 Open Roles

Everlaw's careers page lists 33 open roles across its Oakland, New York, Washington, D.C., and London offices. Eleven sit in go-to-market, eight in engineering, four in product and design, four in customer experience, and the rest scatter across business development, finance, operations, and legal.

The engineering roles map directly onto the AI Deep Dive push. Two positions stand out.

Staff/Principal AI Engineer — the most senior AI-specific role on the board. This person owns model integration, prompt infrastructure, and the retrieval pipeline that powers features like Deep Dive. The "Staff/Principal" level signals Everlaw wants someone who has shipped production ML systems before, not a researcher experimenting on the side.

Senior Software Engineer, AI Platform — a layer below, focused on the APIs, data pipelines, and serving infrastructure that let the AI features run at scale across legal datasets that routinely hit millions of documents.

The rest of the engineering bench is traditional platform work: Senior Software Engineer, Database Systems; Senior Software Engineer, Automation & Developer Tools; Security Engineer; and Senior Engineering Manager, Platform Engineering (Developer Experience). These aren't AI roles, but they matter. An AI-native e-discovery tool that can't handle document ingestion at scale or pass a law firm's security audit is dead on arrival.

On the product side, Everlaw is hiring a Principal Product Designer, a Product Design Manager, a Senior Product Designer, and a Senior Manager, Product Lead, all in Oakland. Four simultaneous product-design hires is a lot for a company Everlaw's size. It suggests the design team is building out a dedicated AI interaction layer, not just adding features to the existing interface. Legal professionals don't just need AI outputs; they need to understand why the system surfaced a particular document, challenge its reasoning, and explain it to a judge. That's a design problem as much as an engineering one.

The go-to-market hiring is where the revenue story lives. Everlaw posted three Field Enablement Partner roles (Oakland, New York, D.C.) and two GTM Knowledge Engineer roles in the past seven days alone. These are the people who translate technical capabilities into language that law firm partners and government agency CIOs understand. Hiring them in parallel with the AI engineers means Everlaw is building the sales motion for Deep Dive while the product is still shipping, not after.

Three Pre-Sales Solutions Architect II roles, one per office, reinforce that pattern. These are the technical sellers who run demos, handle proof-of-concept engagements, and answer the "can this actually work on our data?" question that kills enterprise legal-tech deals.

Two customer success roles titled Senior Manager, Customer Success (Enterprise Legal) round out the picture. Retention in legal tech depends on whether firms actually adopt the AI features they bought. These managers are the early-warning system.

What's missing from the board is just as telling. There are no junior engineering roles, no AI research positions, and no data science postings. Everlaw is buying talent that can ship, not training it. The bet is that experienced engineers and product designers from adjacent industries (enterprise SaaS, cloud infrastructure, maybe defense or finance) can learn the legal domain on the job. The legal domain expertise, meanwhile, lives in the customer success and solutions architect roles, where people who already know e-discovery workflows can bridge the gap between what the AI does and what a litigator needs.

Allison Patrick and the Government Land Grab

Everlaw's hiring surge isn't just an engineering story. The company's executive moves in 2025 reveal where it's placing its biggest bets, and the signal is aimed squarely at government buyers.

Allison Patrick joined Everlaw as vice president of sales for the public sector in September 2025, bringing 25 years of experience selling into federal, state, and local government markets. Before Everlaw, she was chief growth officer at IEM, a defense and homeland security contractor. Earlier, she ran DHS programs at SAIC and held leadership roles at Maximus Federal, Accenture, and SRA, the latter now part of General Dynamics Information Technology. Her resume is a map of the government contracting ecosystem.

Patrick's remit covers the full public-sector sales motion: federal civilian, defense, state and local government, and education. She confirmed the hire in a LinkedIn post, calling Everlaw's mission — "to promote justice by illuminating truth" — and pointing to the company's push to transform how agencies handle discovery and litigation.

The timing tracks with a product launch. In early 2025, Everlaw introduced Everlaw Prime, a case-management portal built specifically for Freedom of Information Act and public records requests. The tool gives agencies a single environment for intake, review, and redaction, replacing the patchwork of disconnected systems most FOIA offices rely on. Pauline Day, Everlaw's product director, said the goal was to let teams "focus on the work itself instead of managing the gaps between tools." Patrick's hire gives that product a sales leader who already knows how those agencies buy.

She's not the only signal. Everlaw's LinkedIn activity in recent months shows a string of enablement and go-to-market roles opening up alongside the engineering positions. The company posted a Field Enablement Partner role for its Oakland office — a position Patrick herself shared on her feed — focused on moving public-sector training from "generic skill development into high-stakes performance consulting." A GTM Knowledge Engineer role, also based in Oakland and New York, rounds out the go-to-market buildout.

The pattern is clear: Everlaw is staffing for a public-sector land grab. The AI Deep Dive feature gets the headlines, but Patrick's hire and the enablement layer being built around her tell you where the revenue strategy actually points. Government agencies are the growth vector, and Everlaw just put a career govcon sales leader in charge of it.

Why Legal AI Became a Talent War

Everlaw's hiring push is not happening in a vacuum. Law firms and legal-tech platforms are competing for the same narrow pool of engineers who understand both machine learning and the mechanics of litigation. Bloomberg Law's August 2025 report, "Artificial Intelligence: The Impact on the Legal Industry," found that generative AI adoption among legal professionals accelerated sharply in 2025, with firms moving tools out of pilot programs and into active case workflows. The report's analysts concluded that the disruption is structural, not experimental.

That shift has a direct consequence for hiring: firms need people who can build and maintain these systems, not just evaluate them. The demand is not for generic ML engineers. It is for people who can connect deposition transcripts, privilege logs, and case chronologies to retrieval pipelines and generative models. That intersection is small, and it is getting smaller as more companies enter the space.

The salary numbers reflect the scarcity. Everlaw's GTM Knowledge Engineer roles sit well above the median for early-to-mid-career legal-tech positions. Field Enablement Partners, priced at $113,000–$182,000, signal that the company is also investing in people who can teach law firms how to use these tools in practice, not just build them.

Bloomberg Law's report frames the moment as a turning point. The five featured analyses argue that AI's impact on legal work is no longer theoretical. Firms that adopted tools for document review and discovery are now looking for platforms that can assist with case strategy and motion drafting. That escalation in scope is what turns a software hiring plan into a talent war — the stakes move from efficiency gains to core legal judgment, and the engineering requirements change accordingly.

Who Else Is Hiring — and What They Want

Everlaw's hiring surge is not happening in isolation. The legal-AI talent market in 2025 is a three-tier fight: well-funded startups like Harvey and Legora are scaling fast, incumbents like Thomson Reuters and LexisNexis are retooling their engineering orgs, and adjacent players like DISCO and Relativity are repositioning around agentic workflows. The result is the tightest market for legal-domain ML engineers and litigation-tech specialists the industry has seen.

At the top of the funding pyramid, Harvey and Legora are the most aggressive hirers. Harvey, now valued at $11 billion after raising $200 million in March, reported $190 million in annual recurring revenue in January and serves more than 100,000 lawyers across 1,300 organizations. The company is expanding its AI agent teams and its embedded legal engineering staff that works directly inside client firms. Legora, valued at $5.55 billion on the back of a $550 million Series D the same month, hit $100 million ARR in roughly 18 months from general availability and has grown to 400 employees across nine offices. It acquired Vancouver-based Walter AI in March to add agentic workflow capabilities. Both companies are hiring for the same profile Everlaw wants: engineers who can build on top of large language models and ship product to lawyers who will actually use it.

The incumbents are responding. Thomson Reuters acquired Casetext in 2023 and has since doubled its applied research group to 260 people, folding CoCounsel into a broader agentic platform. It acquired Noetica in February 2026 to add AI-driven transactional analysis. LexisNexis, which posted $1.2 billion in legal-segment revenue in the first half of 2025, has deployed a four-agent system in its Protege General AI product and is building out intelligent tools that its CEO Sean Fitzpatrick said are driving the highest tech spending law firms have ever recorded. Both companies are hiring AI engineers and data scientists, though their job postings tend to emphasize domain authority and citation accuracy over the greenfield agentic work that Harvey and Legora prioritize.

The e-discovery incumbents occupy a different lane. Relativity and DISCO are adding AI features to existing review and production architectures rather than rebuilding from a foundation-model core. DISCO collapsed its suite into a single per-GB fee and is pushing its Cecilia AI agent. Relativity continues to expand its aiR suite. Everlaw, valued at $2 billion in its 2023 Series D, leads the G2 Winter 2026 rankings for the fourth consecutive quarter and is the closest direct competitor to Harvey and Legora in the litigation workflow space.

The talent pool for all three clusters is thin. Harvey's co-founder Gabe Pereyra came from Google DeepMind and Meta; Legora recruited heavily from Stockholm's SSE Business Lab and Y Combinator's winter 2024 batch. Everlaw's new public-sector sales lead signals a push into government and regulated-industry contracts where defensibility and data handling matter more than speed. The companies that win this hiring race will be the ones that can offer engineers the rare combination of working on hard technical problems and seeing their product used by real lawyers on real cases within weeks, not quarters.

The Three-Layer Stack Behind Litigation Engineering

Everlaw's job postings are unusually specific about what they need, and that specificity is the signal. Strip away the benefits copy and the mission statements, and the engineering roles describe a discipline that looks almost nothing like generic software development and only partly like standard ML engineering.

The Staff/Principal AI Engineer posting tells the clearest story. The role demands six years of ML/AI experience and ten-plus years in software engineering, but the actual work described is not "train models." It is: evaluate frontier models from Google, OpenAI, and Anthropic through early access programs; build retrieval-augmented generation pipelines that operate at trillion-token scale across multimodal inputs; design agent systems with supervision, backtracking, and tool use; and work with vector and graph databases for knowledge representation.

This is not prompt engineering. The posting treats large language models as infrastructure to be orchestrated, not tools to be prompted. The engineer's job is to build the retrieval layer, the validation harness, the agent orchestration, and the storage architecture that lets a model reason over hundreds of millions of legal documents without hallucinating a citation that gets a partner sanctioned. That requires someone who understands both distributed systems and legal workflow constraints — the kind of person who has shipped production SaaS systems before and now has to make them defensible in a malpractice context.

The other engineering roles reinforce this. Everlaw is hiring a Senior Software Engineer for AI Platform, a Senior Software Engineer for Database Systems, and a Senior Engineering Manager for Platform Engineering focused on developer experience. These are not AI research roles. They are the infrastructure layer underneath the AI — the systems that make model outputs reliable, fast, and auditable at the scale litigation demands. A single case can involve tens of millions of documents across email, video, audio, and structured records. The platform has to handle that before the model ever sees a token.

The go-to-market roles push the signal further. Everlaw is hiring GTM Knowledge Engineers and Pre-Sales Solutions Architects who need enough technical depth to configure and explain AI-driven discovery workflows to law firm IT teams and government attorneys general offices. These are not salespeople who learned a demo script. They are hybrid roles that require understanding RAG architecture, model evaluation, and legal data formats well enough to answer hard technical questions from sophisticated buyers.

What this adds up to is a picture of litigation engineering as a stack with three distinct layers. At the bottom: platform and database engineers who build the infrastructure that ingests, stores, and indexes massive multimodal datasets. In the middle: AI/ML engineers who design the retrieval, reasoning, and agent layers on top of that infrastructure. At the top: solutions and knowledge engineers who translate the technical capabilities into defensible legal workflows. Most legal-tech hiring focuses on one layer. Everlaw is building all three at once.

The candidate requirements also reveal what the industry has learned the hard way. The emphasis on validation, hallucination risk, and jurisdictional correctness is not theoretical. Law firms have been burned by AI tools that produced plausible-sounding but fabricated case citations, most famously in the Mata v. Avianca case in 2023, where an attorney submitted brief citations generated by ChatGPT that did not exist. Everlaw's postings explicitly prioritize experience with evaluation frameworks, accuracy auditing, and production-grade reliability over raw model-building skill. That is a direct response to the market's experience with AI tools that worked in demo and failed under legal scrutiny.

The high-value skills in legal AI are not in training models from scratch. They are in building the retrieval infrastructure, the validation harnesses, and the agent orchestration layers that make large language models reliable enough to use in a context where a hallucinated clause is not a bug report — it is a malpractice claim. Everlaw is hiring for exactly that, and paying accordingly.

Working in AI? Zero G Talent tracks the openings: browse AI jobs, openings at Everlaw, and the people building the field.