Company : Tuco AI (tuco.ai)

Location : Bangalore, India (remote or hybrid)

Shift : 9:00 AM to 5:00 PM US Eastern Time (about 6:30 PM to 2:30 AM IST during US daylight saving time, and 7:30 PM to 3:30 AM IST the rest of the year)

Type : Full-time

Experience : 2 to 5 years

About Tuco :

Tuco helps sales teams reach leads over iMessage, SMS and email, with AI agents that send messages, handle replies and book meetings. Tuco connects to CRMs such as HubSpot, GoHighLevel, Pipedrive and Close, so customer conversations, contacts and activity stay in sync.

Our customers are mostly in the US. When something breaks during their working day, a message does not get sent, a lead does not sync, or a reply gets missed. Each of these costs the customer a deal. This role exists so that someone is watching production during US business hours and fixes problems before customers notice them.

What You Will Do :

1. Production incidents (about 50% of the role) :

- Own production health during US business hours. You are the first responder for alerts and customer-reported issues.

- Triage each issue : is it real, how many customers it affects, and how urgent it is.

- Find the root cause using logs, database records and the code. Do not stop at the first error message you see.

- Fix the problem, or hand it to the right engineer with a written reproduction and evidence.

- Write a short incident note after each one: what happened, why, how it was fixed, and what will stop it from happening again.

2. Typical problems you will debug :

- Messages that failed, were sent twice, or went to the wrong channel or number.

- CRM sync issues : missing fields, contacts overwritten with blank values, duplicate leads, webhooks that arrive but are not processed.

- Background jobs that are stuck, retrying forever, or silently dropped (Redis / BullMQ queues).

- Frontend crashes, for example a blank page caused by bad data such as an invalid date or a missing field.

- Alerts that fire on things that are not real problems, and real problems that never trigger an alert.

- Database queries that slow down as data grows (missing indexes, full collection scans).

3. QA (about 30% of the role) :

- Test new features and fixes on our staging environment before they go to production.

- Write test cases from real production data and real failure cases, not only the happy path.

- Check that a deployed fix actually works in production, using logs and the database. A passing unit test is not enough.

- Write regression tests so that a bug, once fixed, does not come back.

4. Debugging and tooling (about 20% of the role) :

- Use Claude Code (AI coding assistant) every day to read unfamiliar code, trace bugs across services, write fixes and tests, and run investigations.

- Improve logging and monitoring, so the next incident is faster to diagnose.

- Keep runbooks up to date for common incidents.

Must Have :

- Claude Code : hands-on use for real debugging and coding work, not only for chat. You should know how to give it context, check what it produces, and catch it when it is wrong.

- TypeScript and JavaScript (Node.js) : able to read a large, unfamiliar codebase, follow a request through API routes, services and background workers, and make small, safe fixes.

- MongoDB : write queries and aggregations to investigate data, understand indexes and explain plans, and know how not to damage production data.

- Redis : understand keys, TTLs and caching, and how Redis is used for locks, rate limits and queues.

- BullMQ (or a similar job queue) : understand jobs, retries, delays, stalled jobs, concurrency and dead-letter handling, and be able to tell why a job did not run.

- Debugging method : you form a hypothesis, test it against logs and data, and prove the cause before changing anything.

- Logs : comfortable searching structured logs (Grafana/Loki or similar), and aware that log queries cost money.

- Git and GitHub : branches, pull requests, code review. You never push directly to the main branch.

- Clear written English. Most handovers and incident notes are written.

- Able to work the US Eastern shift consistently.

Good To Have :

- Experience with CRM APIs and webhooks (HubSpot, GoHighLevel, Pipedrive, Salesforce or Close).

- Next.js and React, enough to debug frontend crashes from a stack trace.

- Messaging and telephony APIs (SMS providers, email sending, deliverability).

- Basic macOS administration, SSH and networking.

- Experience with Railway, Vercel or similar hosting platforms.

- Experience with Linear, Jira or similar issue trackers.

- Prior on-call or production support experience at a SaaS startup.

What Good Looks Like In This Role :

- You check the actual data before you conclude anything. "The code says it should work" is not proof.

- You find the root cause instead of silencing the alert.

- You notice when a check, test or query is returning a result that is too clean, and you confirm it really ran.

- You fix the underlying path that caused the problem, not only the one customer who reported it.

- You are careful with production. You read before you write, and you never delete or bulk-update data without confirming first.

- You leave clear notes, so the next person (or the next AI session) can continue without starting over.