Alpaca Senior Software Engineer at Alpaca to design and ship autonomous execution for agents acting on Alpaca accounts. Lead application architecture for safe agent execution and build account-level policy and guardrails for external agent actions.
Responsibilities
We’re looking for a Senior Software Engineer to design and ship autonomous execution for agents acting on Alpaca accounts. The same platform has to serve external algorithmic traders, partners building products on Alpaca, operators, and partners who embed Alpaca so their end users’ agents can act.
This is a hands-on role. You will not manage people directly. You will lead through architecture, code, design reviews, and mentorship, and you will ship the first version yourself. You will partner with Product, Compliance, Risk, and the team that owns identity and authorization to define the control model. Broker risk controls stay with the teams that own them.
Things You Get To Do:
Lead the application architecture for safe agent execution, from the tool call through to our trading systems.
Design and build the account-level policy and guardrails for external agent actions.
Own the services that evaluate those rules on the path into our trading systems.
Define how agent-initiated trading actions behave, with the engineers who own our tools and skills.
Keep Alpaca’s execution model current with the MCP protocol and with connector, plugin, and skill standards, and translate what changes into account policy, guardrails, and trading actions.
Partner with Product, Compliance, Risk, and the identity platform to define the control model and ship it.
Own production reliability for the policy and guardrail services.
Review designs across teams and guide the technical decisions where agent actions meet brokerage systems.
Mentor engineers and raise the quality of design and implementation in this area.
Qualification
Shipped autonomous executionAble to defineProficiency in Python or TypeScriptStrong architectural judgmentClear communication with engineersWho You Might Be (Nice-to-Haves)
Required
5+ years of software engineering experience, building and operating production backend systems. Evidence of scope and ownership matters more than a strict year count.
Shipped autonomous execution: software that took actions under an explicit policy without a person approving each action, and that rejected or stopped actions outside that policy.
Designed and implemented account-level policy in an application: durable rules for what a caller may do, enforced on the action path, with an audit of the decision.
Enough MCP architecture to work with stateless and stateful behavior, and experience building to MCP, or to connector, plugin, or skill standards, as they have changed in production.
Able to define, with Product, Compliance, and Risk, which account rules allow an unattended action.
Proficiency in Python or TypeScript, and a willingness to work primarily in Go.
Strong architectural judgment, with a record of shipping while requirements are still being defined.
Clear communication with engineers, product, compliance, and risk partners.
Who You Might Be (Nice-to-Haves):
Experience with agent identity and authorization, such as OAuth, tokens, scopes, and delegated access, and comfort partnering with the team that owns that platform.
Experience shipping developer-facing tools, SDKs, or MCP servers.
Experience building agents with a framework such as LangChain.
