Careers

AI Engineer

Build the Monastic Media Dashboard with Claude Code for your first three months, then build AI automation for our D2C clients.

Department: Engineering
Type: Full-time
Location: In-office, Surat
Interview: 30 minutes, in person, at our Surat office

Apply with your AI agent Apply on the web form

Monastic Media runs performance marketing and ads videos for D2C brands: ₹40 Cr+ in ad spend managed, ₹175 Cr+ in revenue created for brands, 5+ years. The whole company runs on one product we build ourselves, the Monastic Media Dashboard: leads, calls, WhatsApp, billing, hiring, attendance, payroll and client reporting, on web, iPhone, Android, Mac and Windows. It is built with AI coding agents, every day.

Months 1 to 3: you build the Monastic Media Dashboard with Claude Code. Real requests from the team, shipped to production: React, Vite and TypeScript screens, Postgres tables and row-level security on Supabase, Deno edge functions, crons, and integrations with Meta, WhatsApp, Google, Slack and Zoho.

After that: we take what we built for ourselves to clients. You build AI automation and AI services for D2C brands, such as order and support agents, ad and reporting tools, and internal dashboards.

You work full-time from our office in Piplod, Surat, with the founder. Pay depends on what you have built, not on years. Apply with your own AI agent through our MCP server, one command in Claude Code:
claude mcp add --transport http monastic-media-careers https://api.monastic.media/functions/v1/careers-mcp
Cursor, VS Code and Claude Desktop setup: www.monastic.media/careers/ai-engineer. Or use the web form.

What you will do

Ship features to the Monastic Media Dashboard every week with Claude Code: React and TypeScript screens, Supabase tables, row-level security and edge functions
Write and check the migrations, policies and tests that keep client and team data private
Connect the tools the team works in: Meta Ads, WhatsApp Business, Google Workspace, Slack and Zoho
Build agents and MCP servers that take repeated work off the team: reports, follow-ups, hiring, support
From month 4: scope and build AI automation projects for our D2C clients

What we look for

Real shipped work: a GitHub profile or live links we can open
TypeScript and React in production
Postgres (Supabase or similar): schema design, SQL and row-level security
Hands-on with LLM APIs, tool use, agents and MCP
You use an AI coding agent (Claude Code, Cursor, Codex or similar) every day
Full-time from our office in Surat

Nice to have

Deno or Supabase Edge Functions
Capacitor, React Native or other native app builds
Meta Marketing API, WhatsApp Cloud API or Shopify APIs
You have published an MCP server, a CLI or an open-source package
You did the optional fast-track challenge

Apply with your AI agent

Apply from your terminal.

Our careers page is also an MCP server. Connect your AI coding agent, let it read the role, and apply from where you already work. It can read what we do and the open roles, read this job and its questions, fetch the fast-track challenge, upload your CV, submit your application, send the challenge and tell you where your application stands.

Claude Code

Run this once in your terminal:

claude mcp add --transport http monastic-media-careers https://api.monastic.media/functions/v1/careers-mcp

Cursor

Add this to ~/.cursor/mcp.json:

{
  "mcpServers": {
    "monastic-media-careers": {
      "url": "https://api.monastic.media/functions/v1/careers-mcp"
    }
  }
}

VS Code

Add this to .vscode/mcp.json:

{
  "servers": {
    "monastic-media-careers": {
      "type": "http",
      "url": "https://api.monastic.media/functions/v1/careers-mcp"
    }
  }
}

Claude Desktop and claude.ai

Settings, Connectors, Add custom connector, then paste the server URL:

https://api.monastic.media/functions/v1/careers-mcp

Then ask your agent

Read the AI Engineer role at Monastic Media and help me apply. Ask me before you submit anything.

Your agent sends only what you approve. A CV is required, as a link or a file. Any other MCP client works too: the server URL is https://api.monastic.media/functions/v1/careers-mcp

Prefer a normal form? Apply on the web form

Optional fast track

Creative fatigue flagger.

About 3 hours with Claude Code or your own AI coding agent. A strong submission is a fast track to the in-person interview. Skipping it does not count against you.

We run Meta ads for D2C brands. A creative's click-through rate (CTR) usually falls as the same people see it again and again. A media buyer needs to know which creatives have worn out compared with how each one did in its own first week, not against an account average.

Download the sample data (CSV)

Synthetic sample data, made up for this challenge. Not real ads and not any client's account. One row per creative per day, 1 to 28 Sep 2026. Columns: date, ad_id, creative_name, format, impressions, clicks, spend_inr.

The rules

CTR = clicks / impressions.
Baseline CTR = CTR over a creative's first 7 days with impressions above 0. A day with 0 impressions is a pause, not data.
Current CTR = CTR over the last 7 days in the file.
Drop = 1 - current / baseline.
Flag a creative when the drop is 30% or more (make the threshold a parameter) and it had at least 5,000 impressions in the last 7 days.
A creative with fewer than 14 days of delivery is 'too new to judge', and one under the impressions floor is 'not enough volume'. Neither is flagged.
Explain every flag in one line a media buyer understands.

What to build

A small web app: load the CSV (upload or URL) and list every creative with baseline CTR, current CTR, drop, a status (flagged, ok, too new, not enough volume) and a daily CTR sparkline, sorted by drop.
The same logic as an MCP tool or a JSON API, for example flag_fatigued_creatives(csv_url, drop_threshold), returning the flagged creatives.
Tests for the CTR maths, including a paused creative, a creative that launched late, a creative that recovers, and one with low volume.

What to send

A public GitHub repo
A live URL (any free host: Vercel, Netlify, Cloudflare, Deno Deploy, Supabase)
A 2-minute screen recording (Loom, unlisted YouTube or Google Drive)
The prompts you used and your CLAUDE.md (or AGENTS.md / .cursorrules), committed in the repo

How we judge it

Correct numbers on the edge cases
How you steered the agent: your prompts and CLAUDE.md
Small, readable code that runs the first time
A recording that shows it working from start to finish

Send it

After you apply, call submit_challenge(application_id, email, repo_url, live_url, video_url, prompts_url).

Or paste the four links into the Optional fast-track challenge box on the web form

How we hire

Four steps, one interview.

1Apply with your AI agent or with the web form. Send your CV and links to work you have shipped.
2Optional for the AI Engineer role: do the fast-track challenge, about 3 hours with your coding agent.
3We read every application. If it is a fit, we invite you to a 30-minute interview in person at our office in Surat.
4Then a decision and an offer. The first 3 months are probation.

Where interviews happen

Monastic Media®, 3rd Floor, Stalwart Insignia,
University Airport Road, beside Shantiniketan,
Piplod, Surat, Gujarat 395007