Senior Software Engineer, Agents
New
R
ReplicantConversational AI
Location: Remote
Secondary Locations: Canada, United StatesFull-TimeSenior
Salary not disclosed
Apply NowOpens the employer's application page
Job Details
- Required Skills
- PythonSQLTypeScriptPrompt Engineering
Requirements
- Earn trust quickly with customers, including skeptical stakeholders.
- Translate ambiguous business problems into concrete plans and communicate with technical and non-technical audiences.
- Understand integration patterns for connecting agents to customer APIs, endpoints, and protocols.
- Design clean tool schemas and address authentication, errors, and edge cases so tool calls are reliable.
- Validate bespoke and sometimes under-documented customer APIs before agents depend on them.
- Be comfortable working with external and third-party systems beyond a customer's own stack.
- Shape LLM behavior through structured instructions and examples, and understand why prompts fail.
- Work evaluation-first and apply judgment about what makes a good conversation.
- General-purpose coding experience in TypeScript, Python, or SQL is a plus.
- Prior forward-deployed, solutions engineering, or customer-embedded work is a plus.
- Conversational AI, voice, or contact center domain experience is a plus.
- Experience with LLM evaluation tooling and observability is a plus.
Responsibilities
- Own assigned customer AI agents across discovery, design, build, evaluation, launch, and live operation.
- Map customer workflows and translate business goals and rules into concrete agent scope and success criteria.
- Design and write agent prompts, instructions, and conversation flows, then iterate using evaluations and real transcripts.
- Connect agents to customer CRMs, ticketing systems, internal services, APIs, and other third-party systems.
- Shape tool inputs and outputs, and handle authentication, errors, and edge cases to make tool calls reliable.
- Build test sets covering real cases, unhappy paths, edge cases, and hallucination risks.
- Measure agent quality and harden agents for production, including guardrails, failure modes, latency, and cost.
- Monitor live conversations and tool-call health, debug using logs and transcripts, and ship fixes.
- Feed recurring patterns and missing capabilities back to Product.
View Full Description & ApplyYou'll be redirected to the employer's site
