In a world where products and software can be replicated overnight, intellectual property is one of the last durable moats. Stilta builds agentic AI for intellectual property, giving legal and IP teams software that can reason across massive document sets, surface evidence, and produce litigation-grade work faster and more reliably.
We recently raised a $10.5M seed round led by Andreessen Horowitz, with participation from Y Combinator and operators from companies including OpenAI, Legora, and Lovable. We already work with major IP firms and Fortune 500 companies, and we’re at the very beginning of defining what great AI-native software for intellectual property looks like.
About the role
You'll be one of the first hires outside the founding team - four ex-QuantumBlack engineers, now building Stilta. Your decisions will shape the product for years. You'll build across the full stack: multi-agent reasoning pipelines, retrieval over hundreds of millions of patents and scientific literature, and the surfaces that practitioners at top IP firms and Fortune 500 IP teams use daily. These are problems without established patterns. You're building agentic software that patent lawyers actually trust to do substantive work - and you'll define what this category of product looks like.
What you will doBuild and improve the multi-agent architecture powering invalidity search, infringement analysis, freedom to operate research, and more
Own the full lifecycle of features - from talking to customers about what they need to watching them use it in production
Push code to production daily. Your code reaches users in hours or days, not weeks
Build retrieval, reasoning, and evaluation systems that run autonomously over millions of documents with reliable source-backed results
Set engineering standards for a codebase that will scale to a large team
Exceptional matters more than experienced - we care about what you can do, not how long you've been doing it
Strong engineering fundamentals and opinionated about product
You write software with and for AI agents, and have shipped agentic systems before
You can take a product idea from sketch to production single-handedly - write the specs, build the UI, wire the agent pipeline, deploy the infra
Comfortable across Python and TypeScript, and you've worked with enough infrastructure - vector databases, distributed compute, cloud orchestration - that none of it slows you down
Experience navigating enterprise requirements - SOC 2, data residency, access controls - is a plus
What we can offer
Competitive Salary + Equity
Unlimited Claude Code - experiment with AI‑assisted workflows, prototypes, and automations from day one
A front row seat - you will be shaping the foundation of our business
Stilta are an equal opportunities employer and welcome applications from all qualified candidates, regardless of age, disability, gender, gender identity, sexual orientation, ethnicity, religion, or background. Selection for this role will be based solely on suitability for the position and does not discriminate on any protected characteristic.
Skills Required
- Strong engineering fundamentals and strong product judgment
- Experience writing software with and for AI agents
- Experience shipping agentic systems
- Ability to take a product idea from sketch to production independently, including specifications, UI, agent pipelines, and infrastructure deployment
- Comfort working across Python and TypeScript
- Experience with infrastructure such as vector databases, distributed compute, and cloud orchestration
- Experience navigating enterprise requirements including SOC 2, data residency, and access controls
What We Do
Stilta builds agentic AI software for high-stakes intellectual-property work, beginning with patent litigation. It helps intellectual-property law firms and in-house teams challenge or enforce patents through source-backed, auditable research and analysis. The platform searches and analyzes patents, scientific literature, and web archives to support tasks such as invalidity, infringement, and freedom-to-operate assessments, aiming to make intensive patent work faster while preserving traceability and defensible evidence.
