Intelligence Architect

Basis AI•New York, NY

1d•Onsite

About The Position

Basis builds real agents that do real work in the real economy. Our agents operate for hours at a time, performing end-to-end work for some of the largest accounting firms in the world. We recently raised $100M at >$1B valuation and are racing to deploy the most advanced applied ML at production scale. The Role: Every team building agents has access to the same models. Most of the attention goes to the product and its behaviors. The compounding work sits underneath: what the agent knows, how it reasons about its work, and how we measure whether it did the job well. At Basis, our agents autonomously perform real accounting work: reconciliations, tax preparation, audit procedures, financial analysis. Often, the quality of this work is not determined by which model we use. It is determined by the context those agents receive: the procedures that define what to do, the domain knowledge that defines how, and the evaluations that tell us whether it worked. To deliver frontier intelligence, we treat context like a product. The procedures, the skills, the system prompts. Each is owned, versioned, and held to a quality bar. Intelligence Architects build the coherent systems that construct our agents’ intelligence from disparate information. You hold the mental model of what each agent knows, how it reasons, and how we measure whether it's working. You write the procedures and skills. You define what "good" means at every layer and design how we measure it. Ultimately, you're orchestrating an intelligence whose ceiling rises with every model release, and whose capability compounds with every edit you make.

Requirements

Writing as a craft. You treat language as load-bearing. You catch the sentence that will be misread before it ships. You have excellent taste.
Reasoning about systems. You see how individual parts contribute to an overall system. When something breaks, your first question isn't what failed but how the system allowed it.
Working from causes, not effects. You don't accept "it works" as evidence that you understand why. You want to know what’s caused the improvement, and which layer allowed the failure.
Drawn to the frontier. The questions in this role don't have answers in any textbook. That's the part that excites you, not the part that worries you.
Motivation by learning. You build accurate mental models of new fields fast. The accountants and engineers you work with should feel that you understand their work, not just talk about it.
Comfort in ambiguity. There are often no right answers. What you write becomes binding and shapes thousands of agent decisions. You take that seriously without being paralyzed by it.
Holding many things at once. You hold the user's problem, the agent's behavior, the procedure's language, and the measure of success in your head at once. You don't lose the thread.

Nice To Haves

Engineers and PMs who've drifted from building systems to questioning whether the system is doing the right thing. You debug problems for users that aren't your job to fix.
Philosophers who've spent years writing precise prose about systems no one else understood. The novelty doesn't intimidate you; it's the reason you do the work.
Lawyers who write the kind of language other people have to follow. You've felt the difference between a phrase that holds and one that doesn't.
Accountants who appreciate language and want to shape the future of your profession.
Editors and writers who treat structured prose as engineering. You've seen instructions get misread at scale and you know why.
Academics tired of producing work that nobody acts on. You want your writing to change something, and you want to be where the change is happening.
Anyone who reads about what we're doing and recognizes the work as theirs.

Responsibilities

Architecting and authoring our agents' context. You decide what context our agents need, how it's structured, and what each layer says. You write the procedures, skills, and system prompts they consume.
Setting the standards. Ensuring consistency across our agents is your primary mission. You define how procedures should be written, what makes a skill document load-bearing, when context is ready to ship. Each product team operates to the standards you set.
Signal to action. You watch the agents at work. When their output isn't good enough, you’ll define what's missing in the context and solve it. The measurement work is shared with product and eval teams; the writing work is yours.
Shaping what comes next. Your perspective informs what we build next. You see across both the agents of today and the possibilities of tomorrow as the substrate improves. That view feeds into product decisions about which surfaces become possible and which existing ones graduate to higher capability.