Job description

About the Company Our client is a pre-seed AI startup in San Francisco 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. About the Role This role builds the layer that turns model capability into systems that actually work for users. You'll develop the core agent harness (the execution loop, tool-use strategies, context construction, and model-facing experimentation) and iterate on agent behaviors across real customer workflows and long-horizon tasks. A defining stance: the creative step happens once, at authoring, and what executes afterward is deterministic, compiled, type-checked code rather than stochastic tool-chaining. You'll own that boundary between what the model decides at runtime and what ships as code. You'll build and run evals against replicas of real customer environments rather than synthetic tasks, reliable enough to gate a release; analyze production failures and attribute them to the right layer (model, prompt, tool contract, environment state, or retry logic); and extend the computer-use agent to systems with no API, owning the guarantees that let it run in production at all: a fresh microVM per task, credentials injected at session start and never seen by the model, every action recorded, and any session killable mid-run. What You'll Own • 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 • 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 Requirements Must-Have • 4+ years of engineering experience • Experience with agent frameworks or tool-using LLM systems • Strong in Python or TypeScript, comfortable with modern AI tooling • Experience with model evaluation, fine-tuning, or prompt design • Can own systems end-to-end and debug across the stack • Thinks in terms of systems and user outcomes, not just model metrics • Enjoys debugging messy, real-world failures and turning them into improvements • Wants to work in the layer that turns model capability into systems that work for users • Able to work in person in San Francisco, Monday to Friday, at startup intensity Strong candidates may also • 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 survive partial failure (idempotency, resumability, compensating actions) • Have integrated enterprise SaaS APIs (identity providers, directory services, ticketing, ERPs) • Have experience with large, messy datasets or production logs • Have been an early engineer somewhere they also had to talk to customers Key Details • Location: San Francisco, CA (in-person, Monday to Friday) • Employment: Full-time • Level: Senior / Staff • Experience: 4+ years • Salary: $200,000 – $300,000 USD per year • Visa Sponsorship: Available (H-1B, O-1, OPT) Benefits & Perks • Meals in office • Health insurance • Unlimited PTO Interview Process • Culture screen • Second culture screen • Technical interview • One-day onsite • Offer

Requirements

Must-haves

What You'll Own
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
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
4+ years of engineering experience
Experience with agent frameworks or tool-using LLM systems
Strong in Python or TypeScript, comfortable with modern AI tooling
Experience with model evaluation, fine-tuning, or prompt design
Can own systems end-to-end and debug across the stack
Thinks in terms of systems and user outcomes, not just model metrics
Enjoys debugging messy, real-world failures and turning them into improvements
Wants to work in the layer that turns model capability into systems that work for users
Able to work in person in San Francisco, Monday to Friday, at startup intensity
Strong candidates may also
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 survive partial failure (idempotency, resumability, compensating actions)
Have integrated enterprise SaaS APIs (identity providers, directory services, ticketing, ERPs)
Have experience with large, messy datasets or production logs

Nice-to-haves

None flagged.

Application prep

Use the role’s language and application details while tailoring your materials.

Keywords to mirror

aiapillmpythontypescript

Platform: manatal
Effort: medium

Benefits

Benefits detected in the listing.

Health Insurance

Pto