Founding Engineer, Applied AI & Grounded Answers

The Denominator · Remote (Preferred: San Francisco, Seattle, New York, Austin or Boston) · Full-time

We believe tomorrow will be better than today, and we build with that conviction. We have a small team and a large thesis, and we intend to define the category.

If you don’t think you meet all of the criteria below but are still interested in the job, please apply. Nobody checks every box — we’re looking for candidates who are particularly strong in a few areas, and have genuine interest and capability in others.

About the Role

The Denominator is building from first principles at the intersection of retrieval, AI assistants and the agentic enterprise. We are looking for a founding engineer to own our answer system end to end: a service that answers executives’ questions only from a curated, sourced and growing body of content, cites and dates every claim, and says plainly when it does not know. You will own retrieval quality, the evaluation harness, the identity layer and the integrations that bring answers to where executives work, and every trade-off between latency, cost and correctness in between.

What You’ll Do

Build and own a grounded question-answering service (ingestion, embeddings, hybrid retrieval and reranking, and generation with large language models) in which every answer carries its sources and their dates
Make the system know what it does not know: out-of-scope detection, calibrated abstention, and refusal rules for questions it must never answer, enforced in code and tested on every release
Build the evaluation harness that decides whether we ship: golden question sets, citation-accuracy and faithfulness checks, regression tests on every content and model change, and a human review loop wired into the release process
Bring the service to the AI assistants and enterprise platforms executives already use: standards-based integrations, enterprise sign-in, marketplace review, and response-time and availability targets you set and meet
Own passwordless identity and consent on a managed provider: business-domain validation, rate limits and fair-use controls, with minimal data collection as a design constraint
Design the event model behind every question (topic classification, source and surface attribution, consent state) and the pipeline that keeps it reliable, queryable and auditable
Own the unit economics of every answer through caching, prepared answers, model routing and retrieval-first design

What You’ll Bring

The Toolkit

5+ years of software engineering, including at least two shipping LLM-backed features (retrieval, structured outputs, tool use) to production users and maintaining them after launch
Deep TypeScript and/or Python; production experience with a modern server-rendered web framework on serverless infrastructure
Hands-on experience with retrieval systems: embeddings, vector databases, hybrid search and reranking, and the trade-offs among them
You have built LLM evaluation, not just run it: test sets, graders, citation and faithfulness metrics, and release gates that block bad changes
Working knowledge of OAuth 2.0 and 2.1, OIDC and enterprise identity and SSO: you know what a token carries and what you should not ask for
Experience building integrations for AI assistants and their app ecosystems, or the skill to be shipping on them within a quarter
Performance and reliability instincts: you measure p50 and p95 latency, set SLOs, and design for cost per request from the first commit

The Denominator DNA

You treat “I don’t know” as a feature and a wrong citation as an outage
You care how an answer reads to an executive, not just whether it retrieves
You know models well enough to know where they fail, and you design around it
You are comfortable being the only person who owns a system, and you build and document accordingly
You believe answers can be both fast and trustworthy, and you build to prove it

Preferred Qualifications

Experience building search or answer products over editorial, legal, financial or research content, where provenance and dates matter
Familiarity with the economics of enterprise AI (token costs, model pricing, FinOps), or the curiosity to be fluent in a month
Experience publishing to an enterprise app marketplace or directory with a security review
Familiarity with privacy and consent regimes (GDPR, CCPA) as they apply to identity and analytics data
Prior experience at a company with fewer than 20 engineers; you have shipped systems without a platform team underneath you

Working at The Denominator

Board-level impact from day one: your work reaches the CIOs, CFOs and boards making AI decisions within months, not years
Fully remote, with a preference for San Francisco, Seattle, New York, Austin or Boston; the right candidate can be located anywhere
A direct line to the decisions: no layers between your judgment and the product

Compensation

$180,000 – $230,000 • Offers Equity • Offers Bonus

Compensation decisions are based on various factors, including, but not limited to, experience, skills, certifications, location and business needs. Employees may be eligible for additional benefits, including health, dental and vision insurance, a 401(k), paid holidays and paid time off, and professional development opportunities and tools.

About The Denominator

The Denominator is building the intelligence company for the AI economy. Enterprises are committing trillions to AI, and the executives accountable for that capital must move at pace, keep their options open and answer for the return. We build the empirical intelligence that lets them decide with conviction, preserve optionality in every technology choice and deploy capital with discipline. Learn more at thedenominator.ai.

To apply, write to careers@thedenominator.ai with the role in the subject line.

The Denominator is an equal opportunity employer. We consider all qualified applicants without regard to race, color, religion, sex, sexual orientation, gender identity, national origin, age, disability, veteran status or any other characteristic protected by law.