Init Intelligence — Senior​/Applied AI Engineer, Agent Harness

Init Intelligence — Senior/Staff Applied AI Engineer, Agent Harness

Type: Full-time | On-site | San Francisco, CA
Compensation: $200,000–$300,000 + 1%–2% equity
Hiring count: 1
Visa sponsorship: Yes — H-1B, O-1, OPT
Reports to: Founding team (Isaiah, co-founder, runs the culture screen) — Linked In not provided

Init is building AI coworkers for IT teams: a security-focused product where an agent registers as a governed identity in a customer's directory, requests scoped access per task, escalates for human approval, and drives systems it was never given an API for. Init raised a $6M seed round (no product, no customers at the time), backed by Max Altman and Ben Braverman (CRO of Flexport, which sold for $2B), plus a bench of operator angels across IT and security.

Several design partners today; founding team of three.

Founded: 2026 | Team size: 1–10 (founding team of
3) | Total funding: $6M seed Industry: AI Tools — AI coworkers for ITWebsite: https://init.inc Office:
San Francisco, CA

Note:

the company card on the role page lists the stage as "Pre‑seed," while the About text describes a $6M seed round
. Carried the seed figure from the fuller About text — confirm with Contrario if the stage matters.

• Ground-floor ownership: One of the first engineers, owning the core agent harness end-to-end — execution loop, tool-use strategies, context construction, evals, and the runtime-vs-compiled boundary.
• Serious backing for a seed team: $6M seed backed by Max Altman and Flexport's ex-CRO Ben Braverman, plus operator angels across IT and security; several design partners already live.
• Hard, real problems: Evals against replicas of real customer environments, per-task microVMs, model-blind credentials, and computer-use on API-less systems — production rigor, not demos.
• Comp + equity: $200K–$300K base with a meaningful 1%–2% equity stake at the seed stage.

• No intake-call transcript was included on the role page. An Intake Video is present but has no accompanying text. Update this section once transcript notes are available.

This role builds the layer that turns model capability into systems that actually work for users. The engineer develops the core agent harness (execution loop, tool-use strategies, context construction, model-facing experimentation) and iterates on agent behaviors across real customer workflows and long-horizon tasks. Defining stance: the creative step happens once, at authoring, and what executes afterward is deterministic, compiled, type-checked code rather than stochastic tool-chaining — and this engineer owns that boundary between what the model decides at runtime and what ships as code.

• The core agent harness: execution loop, tool-use strategies, context construction, and model-facing experimentation
• Agent behaviors across real customer workflows and long-horizon tasks
• The boundary between runtime model decisions and deterministic, compiled, type-checked execution
• Evals against replicas of real customer environments, reliable enough to gate releases
• Production failure analysis and systematic robustness improvements, attributed by layer (model, prompt, tool contract, environment state, retry logic)
• Extending computer-use to API-less systems, with production guarantees (per-task microVM, model-blind credentials, full action recording, killable sessions)
• Feedback loops and data systems that get better real-task data into eval and training

• Have built computer-use or browser-automation agents
• Have experience with virtualization and sandboxed execution environments, and scaling them
• Have done AI research and published at top conferences
• Have shipped something where correctness had to…