Software Engineering, AI Systems (SMTS)

Leadout Capital; San Francisco (CA); On-site; USD 180,000 - 240,000; Full time

Job summary

Salesforce AI Labs in San Francisco seeks an experienced engineer to own substantial backend systems for agent development. You will design, deploy, and operate systems that blend AI reasoning with deterministic execution, ensuring reliability and scalable performance.

You will work closely with customers and cross-functional teams to shape product direction, drive early adoption, and evolve the offering toward general availability at Salesforce scale.

Qualifications

• Hands-on experience with LLM-powered systems and production deployments.
• Ability to design for concurrency, state, retries, and partial failures.
• Proven track record shipping backend services from design to production.

Responsibilities

• Build systems that learn from demonstrations and combine AI reasoning with deterministic execution.
• Develop recovery and verification capabilities for adaptation and human intervention.
• Maintain backend services with robust workflow state management and secure access.
• Create evaluations and diagnostics to measure success and explain failures.

Skills

Python

Go

Java

JavaScript/TypeScript

LLM-powered systems

Job description

The Experience

Join a team of 10 within Salesforce AI Labs, bringing computer-use AI into production with the autonomy of a small startup and the reach of Salesforce. We’re building a product that learns from human demonstrations and turns complex workflows into repeatable automation across enterprise applications.

Making computer use dependable raises hard engineering questions. How does an agent know it succeeded? When should it reason, and when should it follow a deterministic path? How should it recover when an application changes or ask for help before making a consequential mistake?

You’ll help answer those questions in production. Our team works directly with customers, building alongside them and turning discoveries into reusable product capabilities. You’ll own substantial systems from design through deployment and help take a new offering from early customer adoption through general availability, backed by Salesforce’s scale and reach.

What You'll Actually Be Doing

Drive Agent Development & Architecture

• Build systems that learn from demonstrations and combine AI reasoning with deterministic execution to complete workflows reliably.

• Develop recovery and verification capabilities so agents can adapt to changes, confirm outcomes, and recognize when human intervention is needed.

• Build backend services with robust workflow state management, failure handling, and secure access to customer environments.

• Create evaluations and diagnostics that measure task success, expose regressions, and make failures understandable.

Build Directly With Customers

• Work alongside customers to understand their workflows, prototype solutions, and validate results in their environments.

• Investigate where automation succeeds or fails, then turn those findings into improvements to the core product.

• Help customers move from their first successful workflow toward repeatable, increasingly self-service adoption.

• Partner with forward-deployed engineers and product management to decide which customer needs should shape the roadmap.

Shape Technical Direction & Grow the Team

• Own technical decisions for substantial components, from initial design through production operation.

• Evaluate model selection, tool use, and context management against reliability, latency, and cost.

• Mentor peers through design discussions and code reviews, and participate in hiring.

You're Our Person If...

• You’ve independently owned a substantial backend service or system through design, deployment, production failures, and subsequent improvements.

• You have hands-on experience building and evaluating LLM-powered systems, and can explain how you assessed their quality and failure modes.

• You can reason through concurrency, persistent state, retries, and partial failures in systems that perform consequential actions.

• You’re proficient in Python, Go, Java, or JavaScript/TypeScript.

• You can turn an unclear customer problem into a bounded technical approach, test it with users, and carry it through to production.

• You want direct customer engagement to be part of your engineering work, including investigating problems and discussing technical tradeoffs.

• You use AI development tools thoughtfully and take responsibility for the correctness, security, and maintainability of what you ship.

Helpful Experience

• Agent tool use, code generation, automated recovery, or evaluation frameworks.

• Browser automation, workflow orchestration, or enterprise integrations.

• Building an early-stage product or working in a customer-facing engineering role.