Founding Engineer

Principal Engineer — TENET Intelligence Arlington, VA / Hybrid with occasional travel

Important Note: U.S. citizenship is required for this role because it is expected to support future U.S. national-security work, including work that may require eligibility to obtain and maintain a U.S. security clearance.

You will be one of two engineers.

TENET produces decision-ready intelligence assessments for Allied governments and global enterprises.

Our customers make foreign-policy and strategic-risk decisions in hours instead of weeks, with every claim sourced and every judgment auditable. We do this with a multi-agent system grounded in formal intelligence tradecraft.

Your works means you will own the platform. Not a feature, microservice, or backend. The full-stack platform.

What you'll actually build

A production system that ingests open, deep, and dark web sources at scale, resolves them into a coherent knowledge layer, applies reasoning via agentic tradecraft constraints, and produces interactive assessments where every sentence is traceable to evidence.

The architectural challenges are real: identity resolution across languages and sources, agent orchestration that holds up under audit, schema evolution that doesn't break historical work, model-agnostic deployment across commercial and sovereign environments.

This requires full-stack experience.

You will build the analytical infrastructure that lets a small team produce work that today requires a six-person intelligence cell two weeks.

Who we're looking for

Three things, non-negotiable.

Integrity. You will see customer questions that reveal sensitive intent. You will work on systems that shape real decisions. We need someone who treats provenance, source protection, and analytical honesty as personal commitments, not boxes to tick.

Intelligence. Architectural taste matters more than years on a CV. We want someone who has shipped non-trivial distributed systems, who reads papers and writes code, who can hold five abstraction layers in their head at once and pick the right one for the problem. Background in any of: data engineering at analytics scale, knowledge graphs and entity resolution, multi-agent LLM systems in production, secure cloud architecture across commercial and sovereign environments.

Drive. You ship. You don't produce architecture diagrams for weeks before writing code. You take a half-formed problem at 9am and have something running by Friday. You enjoy that.

On what code means in 2026

"Lines of code typed per day" is the opposite of what we measure.

The bottleneck is judgment, system design, evaluation, and taste. Scalability, modularity, and simplicity are what I care about. Your job levels up so that you can scale agentic engineering. The best engineers we work with ship five to ten times what they did in 2023 because they spend their time on the parts models still get wrong. If that excites you, talk to us.

If the idea of working alongside agentic tooling makes you defensive, we're not the right fit for each other.

What you get

Working on human-centric software that contributes to national security. This is the mission. We help our country and its Allies by providing finished, vetted intelligence at the speed of the threat.

Direct work with the founder on architectural decisions. A real seat at the table, not a Jira queue. A mission with weight: the customers tell us the work matters.

The obligation to dissent. I do not want you to execute what you hear and see. You will be a member of the human team at TENET, which means that your ability to think critically and push back when you disagree is the reason we hired you.

Founding-engineer equity (1-2%) and competitive compensation ($150-225k) for senior engineers in this space.

How to apply

What matters more than your resume is sending a 200-word note to contact@tenetintel.com on one of these:

A. How would you resolve "John Xu" the supply-chain executive from "John Xu" the academic researcher across forty languages and two hundred sources, and how would you know when you're wrong?

B. How would you make a multi-agent reasoning chain produce the same answer twice when you run it twice, and what would you refuse to make deterministic?

C. How would you deploy the same analytical platform to an allied MoD on-prem environment and to a U.S. commercial cloud customer without maintaining two codebases?

Tell me how you'd start, what you'd build first, and what you'd refuse to build. The note tells us more than your CV will.