SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
6sense is building the contextual intelligence layer for B2B, with agents and MCP (Model Context Protocol) servers already in production serving real customers. This role extends that foundation by building more capable agents, strengthening the API and tool layer, and creating an ecosystem for external agents, clients, and partners to connect.
You will work across the full journey from a customer's agent to 6sense data and actions, including APIs, tools, MCP, orchestration, context, evaluation, and security. You'll write production Python daily, own multi-month projects, support teams and customers who depend on these systems, and keep the platform current as the agent tooling landscape evolves.
Key ownership areas:
**Agents and Orchestration:** Extend LangGraph orchestrators with typed state, conditional routing, subgraphs, checkpointing, durable execution, and human-in-the-loop steps. Choose among routing, orchestrator-worker, plan-and-execute, reflection, and multi-agent handoff patterns. Run agents headless—API- and MCP-first—across in-app, chat, workflow, and partner surfaces. Maintain reusable components for state, memory, tool-calling, and agent-to-agent delegation with scoped permissions.
**APIs, Tools, and MCP:** Evolve APIs and tools agents use to read 6sense data and take action. Tune tool design for agent success (names, descriptions, schemas, response shapes, error messages, token-efficient outputs). Redesign and scale the MCP layer for production workloads including concurrency, per-tenant quotas, latency budgets, caching, and graceful degradation. Onboard external MCP clients, AI assistants, agent platforms, and partners. Harden execution with idempotency, retries, rate limits, permissions, and spend controls.
**Context and Evaluation:** Partner with the Contextual Intelligence team to expose context to agents through MCP, APIs, and tools. Improve permission-aware retrieval and grounding through chunking, hybrid search, reranking, and citations. Extend evaluation using golden datasets, tool-selection accuracy, end-to-end task success, and regression suites. Expand tracing and usage analytics.
**Security, Privacy, and Trust:** Make security a first consideration in every tool, API, agent, and MCP change. Enforce tenant isolation and user-level authorization. Defend against prompt injection, tool poisoning, excessive agency, and data leakage. Set runtime limits on steps, tokens, time, and cost.
**Customers and Engineering Craft:** Triage and debug customer-reported issues, identify root causes, and deliver fixes. Participate in customer and partner technical calls. Write design documents, review code and designs, mentor engineers, and participate in on-call rotations. Track developments in MCP, agent frameworks, and model capabilities.
This is not an architect role (you'll set direction through working code and design documents), not MCP-only (MCP sits alongside APIs, tools, and orchestration), and not greenfield (these systems are live with real customers depending on them).
**Requirements:**
- 5+ years of experience building and running production software, with strong distributed-systems fundamentals (idempotency, fault tolerance, backpressure, failure recovery)
- 2+ years of experience building production LLM-based systems, including agents used by real customers
- Strong Python skills and production experience with LangGraph or equivalent orchestration framework
- Experience designing APIs and tool interfaces for LLM callers (schemas, descriptions, error handling, tool-selection behavior); direct MCP experience strongly preferred
- Experience scaling an API or platform used by external customers or partners (authentication, rate limiting, versioning, backward compatibility)
- Production experience with retrieval and grounding, as well as writing and running LLM evaluations
- Experience with multi-tenant, security-sensitive cloud systems, preferably AWS (authorization boundaries, data-leakage prevention)
- Experience debugging production issues across services and supporting customers directly or through support teams
- Track record of owning multi-month, cross-team projects independently and communicating clearly with technical and non-technical audiences
- Bachelor's degree in Computer Science, Engineering, or equivalent practical experience
**Nice to Have:**
- Experience building or operating MCP servers in production, or contributing to MCP open-source projects
- Experience serving context to agents from signals, entity data, knowledge graphs, or enterprise data platforms
- Experience integrating with external agent platforms or AI assistants as a partner
- Experience with B2B go-to-market, RevTech, CRM, or intent-data platforms
- Experience with model routing, caching, and cost/latency optimization
- Experience with Kafka, Snowflake, Databricks, or Spark
