Software Engineering MTS

Salesforce•San Francisco, CA

6d•$117,200 - $194,200•Remote

About The Position

Salesforce is seeking a Software Engineering MTS (Member of Technical Staff) to join the Component Accessibility Team (CAT). This team focuses on improving software accessibility by developing AI tools that assist developers in writing accessible code. The goal is to make accessibility an automated, in-loop part of the development process, rather than a manual, after-the-fact audit. The team builds an accessibility agent that provides accessible code suggestions within the developer's workflow, ensuring WCAG compliance before code reaches production. This role is crucial for business success, as accessibility is increasingly a requirement for enterprise and public-sector contracts, and it helps Salesforce avoid legal and regulatory risks. The CAT accessibility product is already operational at scale, with a PR-review presubmit that has successfully reviewed thousands of pull requests.

Requirements

Strong software engineering fundamentals and a track record of shipping code.
Proficiency in JavaScript/TypeScript and modern web development.
Familiarity with at least one component framework (LWC, Aura, React, Vue, etc.).
A quality mindset, utilizing tests, metrics, and evaluation to prove functionality.
Clear written and verbal communication skills, capable of explaining technical tradeoffs to both engineers and non-experts.

Nice To Haves

Hands-on experience with AI/LLM-assisted development, including building agents, tools, MCP integrations, prompts, or evals (LangSmith or similar eval tooling is a plus).
Working knowledge of web accessibility (WCAG 2.2, ARIA, screen readers, keyboard navigation, focus management, color contrast).
Familiarity with the Salesforce UI ecosystem (LWC, Aura, Base Components) and testing frameworks (Jest, Selenium/functional tests).
Experience with developer-tooling adoption, telemetry, or metrics-driven product work.

Responsibilities

Build and evaluate AI-powered accessibility tools, including fix tools and agent skills for generating, validating, and repairing accessible code.
Design and expand the evaluation dataset and reviewer set to prove correctness per WCAG criterion with measurable data.
Treat evaluation as a first-class deliverable, ensuring no AI capability ships without a measured quality bar.
Encode WCAG success criteria and Salesforce internal standards into the agent's knowledge base for correct code suggestions.
Expand per-criterion coverage and grow the reviewer set to bridge the gap between detection and fixing of accessibility issues.
Package tools into Claude surfaces developers use, removing friction to make accessibility automatic.
Remediate real Sa11y test failures and feed learnings back into the tools.