- Salary
- $100 – $160/yr
- Location
- San Francisco, California
- Type
- Full-time
- Department
- Engineering
- Source
- Y Combinator
Description
Build the deterministic AI execution layer for enterprise operations
> Our Thesis: While consumer AI tools and coding assistants (like Claude Code) focus on unconstrained, single-user tasks, mission-critical enterprise operations require a fundamentally different architecture. Enterprise adoption cannot rely on free-roaming personal agents. It demands deterministic execution engines, robust operational guardrails, automated verification loops, and bulletproof process orchestration.
Enterprise software integrations are still painfully fragile: workflows break silently, APIs fail unpredictably, systems drift out of sync, and teams waste countless hours maintaining infrastructure never built for real-time operations at scale.
Our platform powers operations for supply chains, logistics, healthcare, financial operations, e-commerce, and enterprise sales. When our systems run, AI agents execute business processes deterministically: orders move, invoices sync, warehouses operate, contracts renew, revenue gets recognized, and critical workflows execute in real time.
What You Will Work On in Your First 90 Days
- Deterministic Execution Engines: Build production-grade AI agents designed for structured process execution rather than open-ended chat loops.
- Guardrails & Evaluation Suites: Implement programmatic guardrails, schema enforcement, human-in-the-loop fallback triggers
This Role IS for You If
- You understand why enterprise AI requires strict guardrails, transactional guarantees, and deterministic process execution over unconstrained agents.
- You are a systems-minded full-stack engineer who loves tackling difficult backend, distributed queue, and non-deterministic agent problems.
- You think like a product owner, take end-to-end ownership, and thrive in fast-moving startup environments.
This Role IS NOT for You If
- You think building enterprise AI is just writing prompts or making simple calls to OpenAI/Anthropic APIs.
- You dislike operational complexity or debugging non-deterministic failure modes and state drift.
- You require long, slow timelines or highly structured corporate environments.
Tech Stack & Tooling
Languages & Frameworks: Python, React, Postgres
⚠️ Apply By Answering ⚠️
1. What excites you about building guardrailed, process-executing AI execution engines over open-ended chat assistants? (2–3 sentences max)
2. What is the most technically complex agent workflow, guardrail system, or real-time distributed pipeline you've built? Highlight how you handled state, failure modes, or verification loops.