Handpicked roles at some of the most technically ambitious infrastructure and AI companies in the world. Built in NYC.
Select roles below and scroll to the bottom to express interest.
Serverless infrastructure for AI and GPU workloads so engineers can run inference, training, and batch jobs without managing cloud infrastructure.
Why it's interesting — Greenfield infra work on a platform that's genuinely rethinking how compute is provisioned. Small team, high ownership, real systems problems. Quietly building one of the best engineering teams in NYC.
Fast, cost-efficient vector database built for scale. Designed from scratch for low-latency nearest-neighbor search on large datasets.
Why it's interesting — Work on a deeply technical search and storage engine built from first principles alongside a tiny, elite team. Turbopuffer is scaling fast, but the company will likely stay lean, so this is one of the few chances to join early and have outsized impact.
Frontier research lab building foundation models for spatial and temporal reasoning, trained on billions of action-labeled gameplay clips from Medal — the world's largest gamer moments platform.
Why it's interesting — Unique data moat: billions of action-labeled video clips from real human gameplay, at a scale no synthetic dataset can replicate. The infra challenge of ingesting, labeling, and training on that corpus is genuinely hard.
Open standard and tooling for AI code attribution in Git. Tracks which lines were written by which agent, model, and prompt — surviving rebases, merges, and squashes.
Why it's interesting — Git AI is building the observability layer for the AI software factory, helping engineering teams understand, measure, and improve how coding agents operate in production. Opportunity to join two exited founders with strong product-market fit as a founding engineer.
Secure, elastic infrastructure for running AI-generated code. Sub-90ms sandbox creation, stateful execution, and isolated environments for agents, evals, and code interpreters.
Why it's interesting — The execution layer for the agentic software stack is still being built. Daytona is solving the hard parts — fast cold starts, isolation guarantees, stateful sandboxes — at a moment when demand is accelerating sharply.
Developer platform rebuilding the cloud computing stack. 2M+ users, weekly shipping cadence, fully remote team focused on making infrastructure invisible to developers.
Why it's interesting — Small team with massive reach — 2M+ users and weekly releases. The engineering culture is direct, fast, and serious about craft. Real infrastructure problems at real scale without the overhead of a large org.
AI platform delivering structured expert insights for investment firms and enterprises. Primary intelligence — the kind that moves before public data does.
Why it's interesting — Voice AI is making it possible to automate and parallelize primary research at a scale traditional expert networks never could. Hard problems across AI, data infrastructure, and real-time knowledge synthesis.
High-performance infrastructure for CI and AI coding agents, running tens of millions of Firecracker VMs on its own global bare-metal fleet.
Why it's interesting — Rare chance to work on deep infrastructure problems at real scale: bare metal orchestration, Firecracker VMs, distributed scheduling, and petabyte-scale storage systems. The company is now extending that infrastructure into agent sandboxes and AI coding systems.
Autonomous production engineering platform. Builds a live world model of your stack and deploys specialized agents to diagnose, fix, and prevent production issues — without waiting for a human to page in.
Why it's interesting — Production operations is one of the last high-leverage surfaces AI hasn't fully touched. Antimetal is building the autonomous layer between engineering teams and running systems — a world model that persists, learns, and acts.
Infrastructure for human-AI collaboration in software engineering. Building the control plane for agentic development — making AI code auditable, collaborative, and trustworthy at the team level.
Why it's interesting — Empathic is building the trust and coordination layer for AI coding agents: sandboxing, collaboration, and control as engineering teams adopt increasingly autonomous systems. Rare opportunity to join a deeply technical team as a founding engineer and help shape the company from the ground up.