Company · Careers
Careers at PTAS AI
Agentic AI, document intelligence and finance automation that runs in production. How we work, who we hire and how to apply.
The work
AI that has to be right on a Tuesday afternoon
PTAS AI builds DocPro and FinSuite: agents that read invoices, bank statements and KYC packs, check them against business rules, and post the results to ERPs. The people who use them are accountants and finance operations teams. They will tell you, quickly, when a field is wrong.
That makes the work different from most AI jobs. A model that reads nineteen fields out of twenty correctly is not finished. Somebody has to decide what happens to the twentieth: which fields get a lower confidence score, how the review queue shows them, and what the audit trail records. Much of the engineering here is about that last stretch.
The team is small, so the scope of a role is wide. You will talk to customers, read their documents, ship code that runs in their month-end close and see the result.
How we work
Four habits you would pick up here
Ship to production
The work runs for finance teams who notice when it is wrong. We judge our work by how it holds up in production.
Read the documents
Engineers look at the real invoices, statements and edge cases. Most hard bugs are hiding in some vendor's strange layout.
Own the outcome
You build it, you see how it behaves with customers, and you help fix it. There is no wall to throw work over.
Write it down
Decisions, rules and trade-offs get written up, because an agent can only follow a policy someone has written.
Who we hire
Six kinds of role
These are the role families we hire into. Specific openings come and go, so if your work fits one of these, write to us whether or not a role is advertised.
Engineering
AI and ML engineering
Extraction models, LLM pipelines, confidence scoring and evaluation sets built from real documents.
Engineering
Backend and platform
The agent runtime, APIs, queues and integrations that have to stay up when month-end volume arrives.
Engineering
Document AI
Layouts, tables, handwriting and poor scans. The long tail that separates a demo from a product.
Product
Frontend and product design
Review queues, dashboards and configuration screens used by finance teams all day long.
Delivery
Implementation and finance domain
Mapping a customer's AP, GST or reconciliation process into rules, agents and ERP fields.
Go-to-market
Sales and partnerships
Enterprise conversations grounded in what the product does, with partners who carry it further.
The process
Five steps from application to decision
No puzzles and no trick questions. Each step looks like the job.
How hiring works at PTAS AI
- 01 Apply A short note and your work, by email
- 02 Conversation What you have built and why
- 03 Work sample A real problem, scoped small
- 04 Team session Walk us through your answer
- 05 Decision Yes or no, with the reason
You will know where you stand after every step, including when the answer is no.
How to apply
One email is enough to start.
There is no portal and no cover letter template. Tell us what you have built, what you want to build next, and why PTAS AI.
Email your application-
01
Email connect@ptasai.com
Put Careers and the role family in the subject line, so it reaches the right person.
-
02
Show us your work
A repository, a project, a write-up or a deal you ran. Something you made is worth more to us than a list of tools.
-
03
Tell us what you want next
A few lines on what you want to work on and why here. We read every one.
FAQ
Questions candidates ask
Do you have open roles right now?
Openings change as the products grow. Write to connect@ptasai.com with Careers in the subject even if nothing is advertised. We keep speculative applications on file and come back to them when a matching role opens.
Do I need a finance background?
Not for most engineering roles. You will pick up how invoices, GST and reconciliation work by reading the documents and talking to customers. Implementation roles do need finance process experience.
What is the work sample like?
A small, real problem close to the work, such as extracting a tricky table or designing a review screen. It is scoped to take a few hours, and we discuss your answer with you rather than marking it.
Where would I work?
The registered office is in Narnaul, Haryana. Ask about location and working arrangements for the specific role when you apply, as they vary by team.
Will I hear back if it is a no?
Yes. A person reads every application, and if we are not moving forward we will say so, with the reason when we can give one.
Keep reading
Get to know the work
Join the team
Build AI that finance teams rely on
If you want your work to run in production for real businesses, and to see what happens when it does, we would like to hear from you.