The easy button for employee ownership

What Zolidar is

Seed-stage US fintech. We're building the software layer for transitioning small and mid-sized US businesses into employee ownership. The product surface is broad and the team is small. We ship daily.

The work

You'll move end-to-end across the stack. Some weeks: a Next.js feature from schema through API to UI. Other weeks: a Bedrock agent that needs better retrieval, or a Terraform module that wants rethinking. Our stack today is Next.js / TypeScript, a Python service for financial modeling, AWS + Bedrock, Pinecone, and Postgres + Prisma. You don't need to be an expert in every layer. You shouldn't be uncomfortable in any of them.

This is a build-and-ship role. Not a maintenance role, not a specialist role.

Who you'd be working with

Three of us today. You'd be the fourth.

Ashish Agrawal (CEO) — 11 years at Google building consumer products that reached millions of users, then led VR and AI incubation at Matterport pre-IPO. Bioinformatics research at Los Alamos before that. Engineer by training, Carnegie Mellon — still writes and ships code every day.

Sonali Kothari (COO/CRO) — 7.5 years at Kiva.org, growing it to over $1B in loans as COO/CPO of a multi-sided platform with the partner-driven distribution dynamics we navigate today. Before that, COO of JFFLabs. UC Berkeley and UCLA Anderson.

Yash Bhalodi (Engineer) — based in India. He'd be your closest day-to-day collaborator.

On a team this small, you'd work directly with both founders and your engineering peer every day. No manager, no layers of org chart — every code review, every product call, every architectural debate happens in the same small group.

How we work — and how you should already work

AI-first, specifically. Every engineer here runs Cursor (Ultra plan) and Claude (Max plan) daily, and uses them as a serious extension of their own thinking. We expect this to already be how you work — not a curiosity, but your default loop. If you're still copy-pasting between a free chat tab and an IDE, you'll feel behind from day one.

Beyond the tools:

1. Written-first. Most of the day is async writing — Google Chat, Linear, PR descriptions, design notes. Polish in meetings doesn't substitute for clarity in writing.
2. Frontier-aware. You follow what's shipping in AI tooling because you find it interesting, not because someone made you.
3. Decisions go to the person closest to the problem. We hire for judgment and then let you use it.

Who fits here

We weight, roughly in this order:

1. Motivation and raw cognitive ability over years of experience. A sharp, hungry early-career engineer beats a coasting senior one for this role. Years on a resume is a weak signal; how you think is a strong one.
2. AI-native by habit, not by hype. You can articulate where Cursor and Claude help you, where they fail, and where you override them. You don't ship code you don't understand.
3. Craftsmanship. You sweat the small stuff — edge cases, copy, empty and error states, the details most engineers wave away as good-enough. You want your name on quality work, not just shipped work.
4. Grit. You enjoy hard, unglamorous problems and don't lose interest when the work isn't novel. Most of building a company is unglamorous.
5. Intrinsic motivation to build. Side projects, open source, weird experiments — we want to see evidence. Doesn't have to be at a famous place. Has to be real, and yours.

Who doesn't fit here

We'd rather say this up front than waste your time and ours:

1. You want to work on a high-status, externally validated product. We're building unfashionable infrastructure for a problem most people haven't heard of. We won't be in the tech press.
2. You're treating this as a stepping stone to FAANG. We won't be the resume line that opens those doors, and we won't pretend otherwise.
3. You want a stop-gap. If you're optimizing for "until something better comes along," you'll be unhappy here and so will we.
4. You'd hesitate to show this on LinkedIn. We hire for full commitment, not for parallel rosters.
5. You want tightly-scoped tickets handed to you. The work is fuzzy by design.

What this role builds in you

Engineering judgment that compounds and doesn't go out of fashion.

You'll think critically on live problems every day — not on take-homes, not on tickets pre-chewed by someone else. You'll work end-to-end across product, AI infrastructure, and the deterministic finance underneath, on a team small enough that you can't specialize away from the hard parts.

Taste, comfort with ambiguity, fluency with AI tooling, the ability to take a problem from fuzzy idea to shipped solution without supervision — these outlast specific frameworks and stay valuable wherever you go next.

Where and when you'll work

Location. This role is for engineers based in India, anywhere from a metro to a small town works the same to us. We're not hiring outside India for this role.

Schedule. Your hours are your own, with one exception: a few live calls a week, scheduled within the 9:30 pm – midnight IST window (never Fri/Sat/Sun). The weekly meeting load is light especially if you communicate well asynchronously. We'll walk through specifics in a follow-up conversation.

This works naturally for some people — night owls, people whose lifestyles already align — and it's a real cost for others. We've had past candidates push through it and then quietly resent it. If it doesn't fit your life, please don't talk yourself into it. It won't change.

Nice to have, not required

1. Production LLM / agentic / RAG experience, ideally on Bedrock and Pinecone
2. Terraform and AWS operational depth — IAM, networking, observability
3. A public trail: GitHub, side projects, writing, talks. Showing beats telling.

Email careers@zolidar.com with:

1. Resume, LinkedIn, or portfolio. Whichever shows your work best.
2. Current situation and compensation. Where you're based, what you're doing today (employed / freelancing / studying / between roles), your last compensation, and what you're expecting.
3. One thing you've built using AI dev tools. What you built and which tools you used (Cursor, Claude, Codex, whatever). Five bullets or a short paragraph is enough.

We read every email. Short and thoughtful ones get a real reply.