Senior Systems Architect for Probabilistic AI Systems

We're architecting the shift from deterministic features to probabilistic AI systems. You will own the evaluation frameworks and trust architecture for AI agents processing $1B+ in GMV. You'll manage hypothesis portfolios, not feature backlogs. This is not a "playbook" role.

The Stack & Systems You'll Work With

Table Stakes (Required Fluency)

• Product Strategy: 10+ years shipping products that required inventing new methods, not just optimizing existing ones.
• AI/ML Systems: You understand LLMs, embeddings, and retrieval systems well enough to architect products around them.
• Statistical Analysis: You can design evaluation frameworks and statistical acceptance criteria.
• Technical Fluency: You are hands-on enough to write SQL and prototype with prompts.

Differentiation Skills (Preferred, Not Required)

• API Design: Experience designing APIs that autonomous agents will consume.
• Marketplace Dynamics: Experience in 2-sided marketplaces or complex economic systems.
• Python: Preferred for prototyping and data analysis.
• Hypothesis-Driven Development: A formal background in building and killing initiatives based on rigorous data.

The Constraints You'll Navigate

• You will be managing a probabilistic product; capabilities emerge, they are not scheduled. There is no deterministic roadmap.
• You must balance the AI Decision Triangle: Your product cannot be Fast, Capable, and Cheap all at once. You will own the strategy for which corner to optimize for.
• Non-Zero Marginal Cost: Every action a user takes has a real compute cost. Your unit economics are a primary design constraint, not an afterthought.

Why This Matters (The "Glow")

The commerce economy ($8.1T) is being rewired by AI agents. This creates a fundamental "Crisis of Confidence," as users are forced to trust "black-box" systems. The companies that win will be those that build the "truth layer" first.

Demand.io is building that truth layer. We are a profitable, founder-led core team of ~30 supported by a ~100-person global staff. We facilitate over $1B in ecommerce transactions annually by providing verifiable commerce truth. Our core products are the proof:

• ShopGraph: Our proprietary "truth engine," a knowledge graph that provides the canonical source of truth for 10M+ products.
• SimplyCodes: The AI agent that saved consumers $8B in 2024 by finding and applying the best promo codes at checkout.
• Product.ai: The forthcoming agent that solves the consumer "Crisis of Confidence" by providing deep, expert-level advice.

We have no VC pressure, no politics, and no bureaucracy. We operate as a tight "Special Forces unit" running on our proprietary AI-native OS (AIOS), not a traditional corporate hierarchy.

How We Work (The AIOS Mandate)

This is not a traditional company. Our founder designed a new, proprietary system for Context Engineering from first principles.

We run on AIOS (AI Operating System), our internal knowledge graph and cognitive acceleration system. Your work is not just to manage; it is to participate in this advanced cognitive system. This means you will:

• Architect Grounding Files: You will codify your strategies, evaluation frameworks, and unit economic models into structured markdown. This is how you "teach" our AI agents and compound your knowledge.
• Orchestrate AI Systems: You will act as a conductor, guiding AI to generate, test, and refine outcomes based on the context you've engineered.
• Compound the OS: Every system you ship and every framework you codify makes our core OS smarter, creating compounding leverage for the entire company.

Your Mission (Problems to be Owned)

You will own the metamorphosis of SimplyCodes into agentic infrastructure. This means:

• Leading the strategy for our ChatGPT app, transforming it from an experiment into the dominant checkout companion.
• Architecting the "Trust Layer" for AI-mediated commerce, productizing our verification signals.

You will design the evaluation frameworks that define "good" in our probabilistic systems. This means:

• You will manage hypothesis portfolios and real options, not feature backlogs.
• Your acceptance criteria are the product requirements. You'll own P&L responsibility with veto power over what ships.

You will architect the economic model for our variable-cost AI products. This means:

• Owning the unit economics, from cost-per-query to the lifetime value of an agent-acquired user.
• Shaping ShopGraph's knowledge architecture and guiding Product.ai's emergence strategy.

You will partner directly with our founder-CEO and peer with our VP-level leaders to architect products that can't be specified in advance—only discovered.

Who You Are

This role is for a First-Principles "Systems Architect" who:

• Thrives in ambiguity. You don't need a roadmap. Nebulous, 0-to-1 problems are energizing, not stressful.
• Is a systems thinker. You see feedback loops, not linear processes. You are comfortable killing initiatives when data shows failure.
• Treats uncertainty as data, not risk. You can hold multiple competing hypotheses simultaneously and design experiments to falsify them.
• Is a "Builder Who Leads." You are hands-on enough to write SQL and prompt-engineer prototypes to find the truth yourself, not wait for a report.
• Views constraints (cost, latency, trust) as design primitives. You build from fundamental truths, not borrowed playbooks.

The Anti-Profile (You will hate it here if...)

• You need a quarterly roadmap with committed features.
• You find comfort in Scrum ceremonies, story points, and safety in proven playbooks.
• You believe your job is to "coordinate" layers of stakeholder alignment.
• You are uncomfortable with P&L responsibility or the authority to kill a project.
• You wait for PRDs to be handed to you. Here, your evaluation framework is the PRD.

What You Bring

At Demand.io, resumes, pedigree, and credentials are "irrelevant". We filter for your cognitive identity and your proof-of-work, not your resume.

Minimum (Hard Filter)

• 10+ years of shipping products that required inventing new methods, not just following playbooks.
• A proven track record of building in ambiguous, emerging spaces (e.g., 0-to-1 products, new platforms).
• Deep fluency in data analysis and systems thinking.
• Exceptional written communication. We run on memos and grounding files.
• Based in or willing to relocate to Santa Monica, CA.

Preferred (Signals of "10x")

• Direct experience as a PM for AI, ML, or other probabilistic systems.
• A background in FinOps, economics, or managing variable-COGS products.
• Experience in API-first products designed to be consumed by other systems or agents.
• Early-stage startup experience (<50 people).

A note on AI requirements: We're inventing a new discipline. We don't expect you to be a 10-year expert in a 2-year-old field. We hire for a steep trajectory: 70% excellence in your core craft (systems thinking, product judgment), 20% demonstrated AI experimentation, and 10% raw learning velocity. We will provide the domain; you must provide the hunger.

A note on the "Preferred" list: If you don't check every box but you're a first-principles thinker with 10x potential, we want to hear from you.

The Contract

This is a high-trust partnership, built on radical transparency.

Compensation

• Year 1 Cash Comp: $250,000 - $450,000
• Equity: 20-70 basis points (0.20% - 0.70%) of the company.
• Partners Program (PSB): Our profit-sharing bonus model with performance multipliers that can 5-10x total comp based on your impact.

Location & Work Model

• HQ: Santa Monica, CA
• Work Model: We operate 5 days in-office at our Santa Monica HQ to maximize collaborative velocity, with hybrid flexibility for deep work.
• Relocation: We support relocation to LA for the right candidate.

Benefits & Perks

• Health, Dental, Vision: We cover 100% of premiums for you and all dependents.
• Paid Lunch Daily: We eat together as a team.
• Unlimited PTO: We trust you to manage your time and energy.

Our Application Process

We run a high-signal, diagnostic-first process that respects your time and craft. We've abolished 'Easy Apply' and generic recruiter screens.

We review all diagnostics within 48 hours. The entire process, from L1 submission to offer, can take as little as 1-2 weeks for the right candidate. Velocity is a cultural signal.

To begin, please visit our careers hub.

→ Start Your Application

Amplify This Signal

• Follow our Company Page on LinkedIn to see more roles and content.
• Connect with Michael Quoc (CEO) if this mission resonates with you.

(Connecting with us signals your genuine interest in our mission and what we're building. We prioritize candidates based on mission alignment.)