Skip to content

careers · full-time

Founding AI Engineer.

Two jobs in one seat: build the AI Intelligence Workforce, and own its deployment at one or more client businesses, from first visit to live system.

the seat

You build our product, and you deploy it at our customers. At each client you learn how the business really runs, set our software up on top of their Tally or SAP, connect their data, build the automations they need and keep them working. What you learn there goes back into the product, so the next client gets it faster.

8 hrs → 2 min
order entry, in production
~100 yrs
of operations we rebuilt
Live
agents on real customer data

at a glance

Employment
Full-time · PotionLabs
Location
Park Street, Kolkata (on-site)
Start
Immediate
Openings
2 openings
Hours
Mon–Fri 10:00–18:00, Sat 10:00–14:00
Compensation
₹6–12 LPA, based on skill, plus ESOPs

what you own

Four surfaces, and you are accountable for all of them.

01

Find the problem yourself

  • Sit with a customer's operations, sales and factory teams and watch how the work really gets done.
  • Find the slow, manual and repeated steps — including the ones nobody asked you to fix — and decide which is worth solving first.
  • Do the same for our own product: find what is slow, missing or broken, and raise it.
02

Deploy at the customer

  • Set our software up on top of the client's systems, such as Tally or SAP, and take it live.
  • Turn registers, spreadsheets and paper forms into clean, structured data, and prove the new system matches the old one, row by row.
  • Train the client's team, and own the deployment for one or more clients after go-live.
03

AI that takes real actions

  • Build agents that read data, call tools and update records, with a person approving where it matters.
  • Make them safe to run every day on live data: retries, jobs that are safe to run twice, clear logs.
  • Test every agent with evals — real cases with known right answers — before go-live, and keep measuring after.
04

Build and ship the product

  • Build the AI Intelligence Workforce itself: backend services, APIs, database tables and background jobs.
  • Turn what you learn at one client into a product feature, so the next client gets it without custom work.
  • Put changes live, watch them, fix what breaks, and explain what you built to a factory manager in plain words.

day one, you inherit

  • Real problems inside real client businesses from week one.
  • Full ownership of what you build, from the first conversation to the live system.
  • Equity through an employee stock option plan (ESOP) as a founding team member — amount and vesting agreed with you before any offer.
  • Direct work with the founder, and freedom to use the newest AI tools.

what we need from you

Experience

  • 1–3 years of real building — a job, or serious work that real people used.
  • Two or three meaningful projects you owned from start to finish: a real problem, real users, real data, still working after the first week.
  • Mandatory: something you built that nobody asked you to build, that people still use, with an AI feature that takes real actions. A chat window alone does not count.

Technical skills

  • Strong, working Python that another person can read and change.
  • Good SQL: simple table design, joins and totals on your own.
  • Backend basics: APIs, databases and background jobs.
  • Evals and monitoring: you test AI output against a fixed set of cases and watch it after it ships.
  • Daily use of an AI coding tool — Claude Code, Cursor or Codex — and the habit of checking what it writes.

How you work

  • Entrepreneurial: you pick up meaningful work on your own, without waiting to be told, and see it through to a result.
  • Honest about your work: which parts you built, which a teammate built, and which an AI tool wrote.
  • No chatbot-only projects, tutorial projects presented as products, or tools on your CV that are not in your code.

not required

  • Work with ERP or accounting data, such as Tally or SAP.
  • Systems driven by rules or config files, not one-off code for one customer.
  • Tool calling, RAG, MCP, and prompt and context design for AI agents.
  • Some React or TypeScript.

who you are joining

PotionLabs.

AI that evolves with your company.

Most software makes a business change to fit it. Ours learns how the business already works — from its emails, spreadsheets, registers and the knowledge in people’s heads — and keeps rewriting itself as the business changes. We sit on top of the system a customer already has; their books stay in Tally or SAP.

Our first product, the AI Intelligence Workforce, is live in production at JD Jones & Co., a century-old manufacturer, where order entry went from about 8 hours to about 2 minutes. The headless kernel — the architecture where a new customer is read rather than custom-built for — is in build.

Manufacturing is our proving ground, not our ceiling. The team works from the Park Street office in Kolkata.

how to apply

Send your two or three best projects (code, a live link, or a short write-up of what you owned), plus the thing you built that nobody asked you to build — with its AI feature and who uses it today. In the interview we open your code together and ask you to change it. Degree, college and CGPA do not matter.