Skip to main content
Loading
loadingbar
Loading, Please wait..!!

Senior/Staff Applied AI Engineer, Agent Harness

  • Senior/Staff Applied AI Engineer, Agent Harness job type Posted on: Nov 17, 2026
  • Senior/Staff Applied AI Engineer, Agent Harness experience level Transparent Search Group
  • Senior/Staff Applied AI Engineer, Agent Harness employment type San Francisco, California
  • Senior/Staff Applied AI Engineer, Agent Harness salary Salary: $200,000 per year
  • Senior/Staff Applied AI Engineer, Agent Harness employment type Onsite
  • Senior/Staff Applied AI Engineer, Agent Harness salary Full-time

Point Apply Here APPLY LATER

Curious about compensation?

Explore the historical salary trends, average pay, and estimated compensation for Senior/Staff Applied AI Engineer, Agent Harness roles in California.

View Salary Guide →

Job Title :

Senior/Staff Applied AI Engineer, Agent Harness

Job Type :

Full-Time

Job Location :

San Francisco California United States

Remote :

No

Jobcon Logo Job Description :

Job DescriptionJob DescriptionSenior/Staff Applied AI Engineer, Agent Harness\nCompany: Confidential — AI coworkers for IT teamsLocation: San Francisco, CA - in person, Monday to FridayCompensation: $200,000 - $300,000 base + 1-2% equityEmployment Type: Full-timeVisa Sponsorship: Available (H-1B, O-1, OPT)About Init\nInit is building AI coworkers for IT teams. Its security-focused product is an agent that registers as a governed identity in a customer's directory, requests scoped access for each task, escalates to a human for approval, runs in a fresh microVM with credentials the model never sees, and can operate systems it was never given an API for.\nInit was founded in 2026 by Isaiah De La Fuente and Sazzad Islam. The team comes from Stanford, MIT and Delve, among others.The Role\nThis role builds the layer that turns model capability into systems that actually work for users. You will develop the core agent harness, including the execution loop, tool-use strategies, context construction and model-facing experimentation, and improve agent behavior across real customer workflows and long-horizon tasks.\nA defining principle at Init: the creative step happens once, when a workflow is authored, and what runs afterward is deterministic, compiled, type-checked code rather than open-ended tool-chaining. You will own the boundary between what the model decides at runtime and what ships as code.\nYou will build and run evals against replicas of real customer environments, reliable enough to gate a release. You will analyze production failures and trace them to the right layer (model, prompt, tool contract, environment state or retry logic). And you will extend the computer-use agent to systems with no API, owning the guarantees that let it run in production: a fresh microVM per task, credentials injected at session start and never seen by the model, every action recorded, and any session stoppable mid-run.What You Will Own\n\n The core agent harness: execution loop, tool-use strategies, context construction and model-facing experimentation\n Agent behavior across real customer workflows and long-horizon tasks\n The boundary between runtime model decisions and deterministic, compiled, type-checked execution\n Evals against replicas of real customer environments, reliable enough to gate releases\n Production failure analysis and systematic robustness improvements, traced to the right layer\n Extending computer use to systems with no API, with production guarantees (per-task microVM, model-blind credentials, full action recording, stoppable sessions)\n Feedback loops and data systems that bring better real-task data into evaluation and training\nWhat You Bring\n\n 4+ years of experience\n Experience with agent frameworks or tool-using LLM systems\n Strong Python or TypeScript and comfort with modern AI tooling\n Experience with model evaluation, fine-tuning or prompt design\n The ability to own systems end to end and debug across the stack\n A focus on systems and user outcomes, not just model metrics\n You enjoy debugging messy, real-world failures and turning them into improvements\n Ability to work in person in San Francisco, Monday to Friday, at startup intensity\nNice to Have\n\n Built computer-use or browser-automation agents\n Experience with virtualization and sandboxed execution environments, and scaling them\n AI research published at top conferences\n Shipped systems where correctness had to survive partial failure (idempotency, resumability, compensating actions)\n Integrated enterprise SaaS APIs (identity providers, directory services, ticketing, ERPs)\n Experience with large, messy datasets or production logs\n Been an early engineer somewhere you also talked to customers\nBenefits\nMeals in office, health insurance, unlimited PTO.Interview Process\nCulture screen, second culture screen, technical interview, then a one-day on-site.\nBefore you apply: get a free recruiter-readiness score for your resume at CareerStanding — about a minute, no signup, and your resume stays in your browser.

Jobcon Logo Position Details

Posted:

Nov 17, 2026

Reference Number:

16275_8633d511

Employment:

Full-time

Salary:

$200,000 per year

City:

San Francisco

Job Origin:

APPCAST_CPC

Share this job:

  • linkedin

Jobcon Logo
A job sourcing event
In Dallas Fort Worth
Aug 19, 2017 9am-6pm
All job seekers welcome!

Senior/Staff Applied AI Engineer, Agent Harness    Apply

Click on the below icons to share this job to Linkedin, Twitter!

Job DescriptionJob DescriptionSenior/Staff Applied AI Engineer, Agent Harness\nCompany: Confidential — AI coworkers for IT teamsLocation: San Francisco, CA - in person, Monday to FridayCompensation: $200,000 - $300,000 base + 1-2% equityEmployment Type: Full-timeVisa Sponsorship: Available (H-1B, O-1, OPT)About Init\nInit is building AI coworkers for IT teams. Its security-focused product is an agent that registers as a governed identity in a customer's directory, requests scoped access for each task, escalates to a human for approval, runs in a fresh microVM with credentials the model never sees, and can operate systems it was never given an API for.\nInit was founded in 2026 by Isaiah De La Fuente and Sazzad Islam. The team comes from Stanford, MIT and Delve, among others.The Role\nThis role builds the layer that turns model capability into systems that actually work for users. You will develop the core agent harness, including the execution loop, tool-use strategies, context construction and model-facing experimentation, and improve agent behavior across real customer workflows and long-horizon tasks.\nA defining principle at Init: the creative step happens once, when a workflow is authored, and what runs afterward is deterministic, compiled, type-checked code rather than open-ended tool-chaining. You will own the boundary between what the model decides at runtime and what ships as code.\nYou will build and run evals against replicas of real customer environments, reliable enough to gate a release. You will analyze production failures and trace them to the right layer (model, prompt, tool contract, environment state or retry logic). And you will extend the computer-use agent to systems with no API, owning the guarantees that let it run in production: a fresh microVM per task, credentials injected at session start and never seen by the model, every action recorded, and any session stoppable mid-run.What You Will Own\n\n The core agent harness: execution loop, tool-use strategies, context construction and model-facing experimentation\n Agent behavior across real customer workflows and long-horizon tasks\n The boundary between runtime model decisions and deterministic, compiled, type-checked execution\n Evals against replicas of real customer environments, reliable enough to gate releases\n Production failure analysis and systematic robustness improvements, traced to the right layer\n Extending computer use to systems with no API, with production guarantees (per-task microVM, model-blind credentials, full action recording, stoppable sessions)\n Feedback loops and data systems that bring better real-task data into evaluation and training\nWhat You Bring\n\n 4+ years of experience\n Experience with agent frameworks or tool-using LLM systems\n Strong Python or TypeScript and comfort with modern AI tooling\n Experience with model evaluation, fine-tuning or prompt design\n The ability to own systems end to end and debug across the stack\n A focus on systems and user outcomes, not just model metrics\n You enjoy debugging messy, real-world failures and turning them into improvements\n Ability to work in person in San Francisco, Monday to Friday, at startup intensity\nNice to Have\n\n Built computer-use or browser-automation agents\n Experience with virtualization and sandboxed execution environments, and scaling them\n AI research published at top conferences\n Shipped systems where correctness had to survive partial failure (idempotency, resumability, compensating actions)\n Integrated enterprise SaaS APIs (identity providers, directory services, ticketing, ERPs)\n Experience with large, messy datasets or production logs\n Been an early engineer somewhere you also talked to customers\nBenefits\nMeals in office, health insurance, unlimited PTO.Interview Process\nCulture screen, second culture screen, technical interview, then a one-day on-site.\nBefore you apply: get a free recruiter-readiness score for your resume at CareerStanding — about a minute, no signup, and your resume stays in your browser.

Loading
Please wait..!!