AI Engineer- Founder's Office

Vantrock; Maharashtra; On-site; INR 400,000 - 750,000; Full time

Job summary

Vantrock is seeking a hands-on technologist to own the founder's AI stack end to end. You will build internal tools, design the file architecture for quick retrieval, and ship weekly with real users in mind. You will instrument the office with measurable metrics and ensure the stack aligns with Vantrock Intelligence standards.

This is an on-site role with strict confidentiality and discretion. Candidates should have 3–5 years of software experience, be able to demonstrate a production-ready

Qualifications

• Shipped a production tool, repo, or automation with real users.
• Hands-on LLM application work: retrieval, embeddings, evaluation, agent orchestration, tool calling.
• Strong English writing for board-level communication.
• This role is not remote; onsite work is required.
• Discretion with sensitive land, capital and board information.

Responsibilities

• Own the founder's AI stack end to end, integrating Claude, GenSpark and other tools.
• Build internal tools: meeting-to-memo pipelines, action registers, document intelligence.
• Design and enforce file/folder architecture for the founder's office.
• Ship weekly with tools used by real users, not just demos.
• Instrument the office: define metrics and surface them to the founder.
• Align with Vantrock Intelligence standards across the platform.
• Maintain confidentiality with approved tools and accounts.

Skills

Shipped a production tool

LLM application work

English writing

Onsite presence

Discretion

Tools

LangChain

LlamaIndex

pgvector

Next.js

Job description

Vantrock is an India-focused industrial real assets platform covering warehousing, industrial and plotted development across NCR, MMR, Pune and Bengaluru. It builds on its own balance sheet, is backed by global institutional capital, and is working toward a 10 million sq ft pipeline.The hiring thesis: we hire number twos who want to become number one.

About the Role

The founder runs a 12 to 16 hour day across land, capital, construction and the build of the platform itself. Very little of it is organised and almost none of it is instrumented. Meetings are recorded and never converted. Documents sit in four places. Questions that should take twenty minutes take three days because nobody has built the thing that would answer them.This seat builds that thing. It owns the founder's AI stack end to end, turns it into an operating layer rather than a set of open tabs, and then keeps building as the platform grows. Six months in, the founder's office should run on software that did not exist the month before.

Responsibilities

• Own the founder's AI stack end to end. Claude, GenSpark and everything else, wired into one system with shared memory, standard prompts, and retrieval over the real document base rather than a dozen disconnected tabs.
• Build the internal tools nobody else is going to build: meeting-to-memo pipelines, action registers that chase themselves, document intelligence over land files and transaction papers, and dashboards that pull from Drive, Notion, Gmail and Slack.
• Design and hold the file and folder architecture for the founder's office, so that retrieval works without a human in the loop and nothing is ever looked for twice.
• Ship weekly. Working tools in front of a real user, not specifications and not demos.
• Instrument the office. Decide what should be measured, build the measurement, and put the number where the founder sees it without asking.
• Build to the standards set by Vantrock Intelligence, so that the founder's office runs on the same stack as the business rather than a private one growing beside it.
• Hold the confidentiality line absolutely. Approved tools, approved accounts, and a written rule for what never leaves.

WHAT THIS ROLE IS NOT

• It is not a research seat. Nothing here is evaluated on novelty.
• It is not a product-management seat. You build, you do not write requirements for somebody else to build.
• It is not a seat inside Vantrock Intelligence. That is a separate business with its own products. This seat serves the founder.
• THE HOURS, SAID PLAINLY - The founder works a 12 to 16 hour day. Not everyone wants that, and we do not pretend otherwise. What we will not do is waste it. This seat carries a stated arc: roughly 24 months in the founder's office, then a seat in the business on performance.

Qualifications

• 3 to 5 years building software, with Python or TypeScript to production standard.

Required Skills

• You have personally shipped something with real users. A repository, a deployed tool, an automation running in production at your current firm, or a product of your own. You will link it in your application.
• Hands-on LLM application work: retrieval, embeddings, evaluation, agent orchestration, tool calling or MCP. Name the stack you actually use, whether that is LangChain, LlamaIndex, pgvector, Next.js or something else.
• English prose a board member would read without wincing. Roughly a third of this seat is writing.
• You want to be in the room. This is not a remote engineering seat and it is not a hybrid one.
• Discretion. You will see land transactions, capital conversations and board material in your first week.

Preferred Skills

• Founder's office, chief of staff or internal-tools exposure at a startup
• Automation across Google Workspace, Notion and Slack
• Agents running in production rather than in a demonstration
• Working comfort with video, design or presentation tooling
• Exposure to real estate, infrastructure or financial services

CULTURAL FIT

We're building Vantrock over a decade, not a funding cycle. We're looking for people who treat technology as the default answer rather than a supplement, who want to be co-builders of the platform rather than employees executing instructions, and who are comfortable saying "I don't know" and "I was wrong." If you want a defined role with defined edges, this isn't the right seat. If you want to help build an institution from zero, it might be.AI proficiency is a standing requirement in every seat at Vantrock. You use AI tools daily as force multipliers, and you judge their output rather than forwarding it.