Founding AI Engineer

About Second Chair AI

Most legal AI is a toy. Ours scales law firms without hiring more staff.

Second Chair AI produces finished, litigation-ready legal work product - briefs, motions, discovery responses, citation-verified memoranda that attorneys file in court under their own name. Not summaries. Not suggestions. Not "AI-assisted drafts." Finished work.

This system already runs in production inside a 30-person litigation firm. Real cases. Real deadlines. Real consequences if it's wrong. It has multi-agent review pipelines, automatic citation verification, and quality standards that would make most AI companies uncomfortable.

Now we're turning it into a product that scales associate and paralegal work across thousands of firms. We need the engineer who builds it.

What Second Chair AI Actually Is

Second Chair AI was built by a practicing litigator and Managing Partner of the 5th fastest-growing law firm in America (Inc. 5000). He didn't build this because AI is trendy. He built it because he runs a firm, sees exactly where legal labor breaks down, and decided to optimize it.

The system already:

• Produces litigation-ready drafts - motions, briefs, demand letters, discovery responses, that go to court
• Verifies citations automatically - cross-referencing against actual case law databases, not just pattern-matching
• Uses multi-agent architectures - separate systems for drafting, review, citation checking, and refinement
• Operates under real deadlines - court dates don't move because your pipeline is flaky

This is not a research project exploring whether AI can do legal work. That question is answered. The question now is: how fast can we scale it?

The Role

You are the first engineer. Read that again.

You're not joining a team. You're becoming the team. Every architectural decision, every infrastructure choice, every line of code that ships this year, that's you. The systems you build will be the foundation that everything else runs on as this company scales to hundreds, then thousands of law firms.

You report directly to the founder. No layers. No translation. No product managers reinterpreting what matters. You'll sit with someone who practices law every day and knows exactly what "good enough" means in a courtroom, because he's the one filing the work.

What You'll Actually Build

In your first 90 days:

• Multi-agent orchestration - designing and refining pipelines where specialized agents draft, review, verify, and refine legal work product before it reaches an attorney
• Citation verification systems - reliable pipelines that check every citation against real databases (CourtListener, Justia, and Westlaw APIs) because a hallucinated citation is malpractice
• RAG at scale - integrating firm-specific case files, court rules, local practice standards, and client history into generation pipelines
• Quality and evaluation infrastructure - systems that measure whether output is actually good, not just fluent
• Production infrastructure - monitoring, error handling, cost optimization, and latency management. This system runs under court deadlines. It doesn't get to be down.

Within three months: architecting the platform for external law firms, building tenant isolation, designing the integration layer with legal practice management software, and hiring numerous engineers.

What We Need

Non-Negotiables

• You've built LLM-powered systems that real users depend on - not demos, not hackathon projects, not internal tools that three people use
• You understand LLM failure modes deeply - hallucinations, citation fabrication, reasoning breakdowns, retrieval poisoning, and have battle-tested opinions on mitigation
• You grasp local LLM architecture and structuring for efficient token usage
• You know the difference between moving fast and being reckless. You've shipped things quickly that held up under pressure, and you've seen what happens when others didn't
• You have strong taste, you know when something is "not good enough" before you can articulate why
• You operate well in high-ambiguity, high-ownership environments

Strong Signals

• Multi-agent or iterative reasoning systems
• Evaluation frameworks and QA pipelines for LLM outputs
• Experience across multiple models (OpenAI, Anthropic, local models)
• Legal tech, healthtech, or regulated industry experience
• Management or C-suite experience - VP Eng, CTO, or senior tech lead background
• Early-stage startup experience (employee #1–5)

What This Is Not

• Not a research sandbox - we proved AI can do legal work. You're here to scale it
• Not a slow, process-heavy environment - you will ship meaningful work in your first two weeks
• Not a place to learn AI from scratch - we need someone already deep
• Not a wrapper shop - our systems have real architecture: agents, pipelines, verification layers, evaluation loops
• Not a role where "mostly works" is acceptable - our output goes into court filings

The CTO Path

Let's be direct about what this becomes.

• 3 months: You've rebuilt and scaled the core platform. You're making every technical decision. You're defining the engineering culture.
• 6 months: You're hiring engineers 2, 3, and 4. You own the architecture, the roadmap, and the team.
• 12– 18 months: You're the CTO. You're leading a team, presenting to investors, making strategic technology decisions that shape the company's trajectory.

This path isn't guaranteed - it's earned. But the seat is open and the first engineer has the inside track by definition.

Compensation

• Base salary: $150,000–$200,000
• Equity: meaningful early-stage stake - you're the first engineer and the package reflects that
• Location: Denver hybrid (preferred) or US remote
• Health, dental, and vision benefits
• Direct access to founder - no layers, no middlemen

How to Apply

• Your resume or LinkedIn
• One paragraph on the most complex LLM system you've built, what made it hard, and what you'd do differently now

We review every application personally. If you're the right person, you'll hear from us very quickly.