Software Engineer

Location: In person, Chennai ·

Levels: All levels We're looking for exceptional engineers to join a lean, strong team building the agent platform for enterprise finance. If you want to work across the entire stack, ship to production in minutes, and solve problems that don't have off-the-shelf answers, read on.

About Us At Numero, we're building the agent platform that lets enterprise finance teams deploy production AI workflows in days, not quarters. We're backed by $XM+ in funding by leading VCs and already trusted by leading accounting firms like Connor Group and enterprise customers like Zuora, Medidata, and Medallia. From reconciliations to the long tail of finance work that's never been automated, we're small and opinionated — obsessed with two things: how fast a customer goes from first call to a working agent in production, and how good that agent is once it's there.

Team You'll work directly with founders who are deeply technical: they've built at Google, Meta, and Amazon — including core work on Google's V8 engine and React at Meta. You'll learn from some of the best engineers in the world, up close. Engineering is just five people, skewing senior — and we intend to stay lean. We hire only the best, and everyone here is genuinely a genius. Our team includes:

• People who've founded their own companies and served as CTOs, now joining as founding engineers
• Engineers from Freshworks who've built systems operating at incredible scale
• Engineers who've built video transcoding pipelines from scratch to millions in revenue There are no narrow lanes: everyone works across the entire stack, from DevOps to frontend to AI engineering. Everyone has Claude and Codex subscriptions, and everyone orchestrates agents as part of how they build. On a team this small, your work is visible and your impact is immediate.

Engineering Challenges We're not building another SaaS CRUD app. We're building a platform to deploy AI agents in production — and the hard problems span three areas.

Platform engineering. Building a platform is a different game from building a product. You'll wrestle with the right levels of abstraction, plus reliability and durability at the core. An agent might suspend, wait days for human input, then resume exactly where it left off — simple to state, hard to build. What happens when the agent definition changes while it's waiting? How do you version agents and deployments without breaking in-flight work? These are the questions you'll own.

AI engineering. We're a model router, using different models for different jobs — and deciding when to reach for a VLM vs. an LLM. We build golden datasets and evals for every agent before it ships, because nothing off the shelf works for our problem. You'll build an entire eval platform from scratch and figure out how to orchestrate agents reliably.

Product engineering. Our vision: you orchestrate agents the way you'd orchestrate a team. When your "workers" get stuck, they reach out — they email you a summary of what they've done and ask you to unblock them. So we built an inbox as a core platform primitive, the channel through which agents communicate with users. Accountants do their real thinking in spreadsheets, so we're building a spreadsheet from scratch — one of the genuinely hard problems in frontend engineering. This is product engineering at the primitive level: defining and building the core pieces that don't exist yet. This is what you'll build from day one — the primitives of a category that doesn't exist yet.

Tech Stack You're not expected to have experience with anything listed -- yes, you read that right -- you don't need any relevant experience. We are looking to hire the smartest software engineers and smart software engineers can pick up these tools easily.

• React and Next.js on the frontend
• Node.js and PostgreSQL on the backend
• Redis for caching and coordination
• Vercel AI SDK powering our agent and model-routing layer
• GitHub, Graphite (stacked diffs), and Linear

Culture Think Meta meets Google: we move fast like Meta, with the engineering rigor of Google. "Move fast" isn't a slogan — it shapes our technical choices. We run a single full-stack TypeScript monorepo, because every extra language is overhead (build tooling, less shared code, more friction) that slows us down. We use stacked diffs to keep changes small and shipping, and our CI/CD gets your change to production in 10 minutes. We work in person in Chennai. You'll thrive here if you have a bias for action — you ship, learn, and iterate rather than waiting for perfect. You take ownership end to end and move with urgency without dropping the bar on quality. If you want heavy process and long planning cycles, this won't be your place.

Compensation We aim to be top of market pay — the 95th percentile of total compensation in Chennai, not like a typical startup. You get a strong base, a significant performance bonus, and meaningful equity that refreshes every year. Everything is tied to impact rather than tenure, so the engineers who move the needle keep earning more for it over time.

Perks

• Catered lunch every day
• Unlimited holidays
• Insurance
• A team outing every month Candidate
• Experience building customer-facing products at a software company with a high engineering bar
• Track record of driving complex, end-to-end features (not just incremental improvements and refactors) with visible impact on the product
• Product sensibility: you care about UX, speed, and polish, and you push back when something doesn't feel right
• Comfortable working without heavy PM overhead; you can help shape the problem, not just implement a solution
• High ownership mentality: self-directed, full-lifecycle builder who takes a feature from idea to shipped
• Genuine excitement about AI and agents — you want to build at the frontier of how software gets made
• Experience at a startup or a company with a genuinely high engineering bar