APM - toolkits

Who are we?

Composio builds the AI-native automation layer for software teams: SDKs, runtimes, and platform tooling that let developers compose, evaluate, and operate agentic workflows and AI-powered products. Our systems must be reliable at scale - agents must act correctly, integrations must be robust, and integration needs to be an autonomous, observable function that accelerates engineering rather than slowing it. You'll join a small, high-leverage team building the delivery systems and metrics that give product and support teams visibility and predictable delivery.

This role is based in Bangalore (HSR).

Who is our ideal candidate?

A high-agency operator who is also AI-native, and who wants to grow into product. The role starts ops-heavy: you run the operations that get apps live on Composio, keep the toolkits healthy, and automate as much of that work as you can. Over 6-8 months, once those workflows and automations are running, the role shifts towards owning ambiguous product problems across the platform - triggers, tools, skills, auth, governance, and toolkit-level issues like rate limits and MCP hosting. This is not a traditional PM role with an engineering team to direct. It's closer to founder-mode ownership of a product line, where you solve problems using agents and a small ops team.

You've shipped things on your own - a side project, a tool, a micro-business, or a system you built that is still running - and you've run real operations with quality bars, not just executed inside someone else's process. You're at the edge of AI out of genuine curiosity: you use AI tools daily and can point to specific things you built or automated with them in the last month. You pick work by impact, not by how interesting it is, and you have no ego about manual work that isn't automated yet. You can hold a technical conversation with engineers without a translator, and you're comfortable owning problems that don't have a clear owner yet.

What will they own?

• The toolkit number: how many apps are live on Composio, and the health of the ones already live.

• New apps, end to end: run the operations of getting them live - research, setup, QA, launch.

• Existing toolkits: make them better with the product team, and clear the issues that stop them from working well.

• QA and support for every toolkit you bring live. Once it's live, its quality is your problem.

• The ops playbook: build it, run it, scale it. SOPs, small tools, AI agents before humans - you automate yourself first, and when scaling takes people, hiring them is part of the job.

• The metrics for this function: define them, build them, watch them.

• Product problems, as the ops layer starts running on its own: ambiguous, non-engineering problems across triggers, tools, auth, governance and toolkit-level issues - scoped, owned and shipped by you.
  

What are we looking for?

• Experience: 2-4 years. We care more about how fast you've grown than how senior you are.

• Technical floor: can explain what an API and an integration are, reads API docs cold, writes enough code to build small tools and automations. Not an engineer; infra-level work is not the job.

• AI-native automation instinct: your default answer to repeated work is "how do I never do this manually again" - backed by recent, specific evidence of things you've built or automated with AI.

• Dirty-work appetite: you happily run the manual version today while the automation is being built. This and the automation instinct are both required; we never average them.

• Process maturity: you've designed or owned a pipeline with throughput, SLAs, or quality bars, and you left systems running behind you - not just stories.

• Product instinct: you can take a vague problem, figure out what's actually broken, and decide what to build or fix - without waiting for a spec or a dedicated engineering team.

• Organized ownership: routines, follow-ups, nothing dropped. You communicate clearly, in writing and in person.

• High agency in uncertainty: given a goal with no playbook and no owner above you, you define the problem yourself and start before permission.

• Operating maturity: this role carries a number from week one, so it needs someone who has owned outcomes before. It is not internship-shaped.

We'll pass at any stage on candidates who only want the cool parts of the job, choose work by interestingness over impact, can't explain what an API is, or do the same manual work in month three as month one.

Nice-to-have

• Founder or early-stage experience: you've taken something from zero, or joined early and owned a function.

• Experience with SaaS integrations, API products, or developer-facing products.

• Built things with agents or LLM workflows beyond personal use - tools, experiment logs, side projects other people use.

• An engineering-adjacent path: studied engineering and moved into non-technical work, or a non-CS background with things you built on the side.

• High-throughput ops backgrounds (delivery, marketplace, physical ops) count only alongside the AI-native bar above - on their own they are not this job.
  

Outcomes we expect in the first 90 days

• More toolkits live on Composio - you own the supply number, and it moved.

• Existing toolkits repaired or unblocked without anything breaking technically.

• A v1 ops playbook running: SOPs, automations, and AI agents handling the repeatable parts of onboarding and QA.

• Metrics for the function defined and visible: supply, health of supply, pipeline status.

• What this is not: usage. Usage is a product metric; this role is measured on supply and the health of supply.