MCP Developer (AI Tooling & Integrations) at Instantly
Remote · full time · entry level
About the role
## About the role Instantly is building a new product on top of MCP connectors, CLIs and an AI backend, and needs one engineer to own the MCP layer end to end. This is a builder's role. Instantly wants someone who has shipped their own MCP server, not someone who has only plugged existing ones into Claude or Cursor. You'll be the go-to person for how AI agents talk to the platform: the connectors, the guardrails around them, and the developer tooling that ships with them. You'll work closely with a small, fast product team, with real ownership from week one. **Location:** Remote, work from home · **Employment:** full-time · **Experience:** freshers to ~3 years, no minimum ## About Instantly Instantly is a US-based email outreach platform used by sales and growth teams to find leads, send cold email at scale and manage replies. ## What you will do - Design, build and maintain MCP servers that expose the platform to AI agents (Claude, Cursor and other MCP clients). - Own the MCP section of the product: tool definitions, schemas, auth, versioning and reliability. - Build guardrails around agent actions: permissions, rate limits, input validation and safe defaults. - Ship developer tooling alongside the MCP layer: CLIs, SDKs and the docs that go with them. - Wire LLM-powered features into the backend where they make the product better. - Test connectors against real agent workflows and fix what breaks. ## What you need - Has built at least one MCP server themselves, for any tool or service — a public GitHub repo or demo is ideal - Solid backend skills in TypeScript/Node.js or Python - Hands-on experience building CLIs, SDKs or API integrations - Practical AI experience: calling LLM APIs, tool use / function calling, agent workflows - Ships fast and owns things without being chased ## Nice to have - Built guardrails or safety layers for AI agents. - Open-source contributions in the MCP or AI tooling ecosystem. - Side projects or hackathon builds using Claude, Cursor or similar tools heavily. - Experience with email infrastructure or developer platforms. ## Not the right fit - Deep LLM training / research engineers (model training, fine-tuning at scale) — this role is about tooling, not models. - Candidates who have only used MCP connectors, without building one. Freshers and college builders are very welcome. If you've shipped real MCP work, years of experience don't matter. ## Working model - Full-time, remote — work from home.
Responsibilities
- Design, build and maintain MCP servers that expose the platform to AI agents (Claude, Cursor and other MCP clients).
- Own the MCP section of the product: tool definitions, schemas, auth, versioning and reliability.
- Build guardrails around agent actions: permissions, rate limits, input validation and safe defaults.
- Ship developer tooling alongside the MCP layer: CLIs, SDKs and the docs that go with them.
- Wire LLM-powered features into the backend where they make the product better.
- Test connectors against real agent workflows and fix what breaks.
Requirements
- Has built at least one MCP server themselves, for any tool or service — a public GitHub repo or demo is ideal
- Solid backend skills in TypeScript/Node.js or Python
- Hands-on experience building CLIs, SDKs or API integrations
- Practical AI experience: calling LLM APIs, tool use / function calling, agent workflows
- Ships fast and owns things without being chased
Skills: MCP, TypeScript, Node.js, Python, CLI, SDKs, API integrations, LLM APIs, Function calling, AI agents, Claude, Cursor
Browse more jobs on 100Networks