Senior Backend Engineer

Jobtailor; San Francisco (CA); On-site; USD 140,000 - 210,000; Full time

Job summary

Jobtailor is hiring for a backend/infrastructure leader to build a multi-tenant agent execution sandbox and durable runtime on Temporal. You will develop agent harnesses and SDKs, enabling tool-calling, context management, and orchestration with robust guardrails.

Responsibilities include exposing MCP tooling, ensuring interoperability across runtimes, and implementing observability and audit trails for agent execution.

Qualifications

• End-to-end agentic framework required, from ground up design to deployment.
• Design the agent loop: tool-calling, context mgmt, multi-agent orchestration, eval/guardrails, and agent-specific failure modes.
• 6+ years building backend systems or distributed infra at scale.
• Built SDKs or platforms that other engineers rely on.
• Production use of AI-assisted dev tools across full lifecycle.
• Multi-tenant design with strong isolation, ideally in regulated domains.
• Async/concurrent execution patterns (goroutines, async/await, thread pools).
• Secure API design with strong boundary auth.
• Observability/logging/tracing/audit trails.
• Familiar with agent dev kits and internals, not just APIs.

Responsibilities

• Build a multi-tenant execution sandbox for agents with provable data isolation.
• Develop durable execution backbone on Temporal for resilience of long-running agents.
• Create agent harness and SDKs enabling tool-calling, context management, and model orchestration with guardrails.
• Expose MCP tools and ensure external runtime interoperability.
• Implement observability and audit infrastructure for agent execution.
• Design failure model with error handling, resource limits, and graceful degradation.

Skills

Temporal backend

Multi-tenant architecture

Agentic framework

Asynchronous/concurrent execution

Secure API design

Observability/instrumentation

Audit/compliance

Agent dev kits knowledge

gRPC

Event-driven architecture

Data isolation

Tools

Google ADK

Claude Agent SDK

OpenAI SDK

LangChain

CrewAI

AutoGen

Airflow

Prefect

Kafka

Job description

Key Responsibilities

• Build a multi-tenant execution sandbox for agents with strict, provable data isolation.
• Develop the durable execution backbone on Temporal to ensure long-running agents survive failures, restarts, and retries.
• Create the agent harness and SDKs, enabling tool-calling, context management, model orchestration, and the eval and guardrail loops that keep behavior safe.
• Expose MCP tools and provide external runtime interoperability.
• Implement observability and audit infrastructure for agent execution.
• Design the failure model with error handling, resource limits, and graceful degradation.

Qualifications

• Built an agentic framework or runtime end-to-end, from the ground up, not just used one.
• Designed the agent loop yourself: tool-calling, context management, multi-agent orchestration, eval and guardrail loops, and the failure modes specific to agents.
• 6+ years building backend systems or distributed infrastructure at scale.
• Built SDKs or platforms that other engineers build on top of.
• Production use of AI-assisted dev tools (Claude Code, Codex, Gemini CLI, or similar) across the full lifecycle: spec, build, review, and deploy validation.
• Multi-tenant design with hard tenant isolation, ideally in a regulated space (financial services, healthcare).
• Async and concurrent execution patterns (goroutines, async/await, thread pools).
• Secure API design with authn/auth at the service boundary.
• Observability and instrumentation: logging, tracing, audit trails.
• Preferred skills: Workflow orchestration (Temporal, Airflow, Prefect), especially agentic workflows in production.
• Managed agent runtimes (Claude Managed Agents, Bedrock Agents, Google Agent Builder, Azure AI Services).
• Runtime isolation and sandboxing.
• Deep familiarity with agent dev kits (Google ADK, Claude Agent SDK, OpenAI SDK, LangChain, CrewAI, AutoGen) at the level of their internals, not just their APIs.
• MCP or similar tool and extension systems.
• gRPC or modern RPC.
• Multi-provider key management; secrets and credential rotation.
• Event-driven architecture (streaming, queues, event sourcing; Kafka or similar).
• Knowledge graphs for grounding agents (Neo4j, or Obsidian-style linked stores).
• Audit and compliance in regulated industries (PCI-DSS, SOC 2).