Product Manager at Sarvam AI

Bengaluru, karnataka, India -

Full Time

Start Date

Immediate

Expiry Date

21 Jun, 26

Salary

0.0

Posted On

23 Mar, 26

Experience

5 year(s) or above

Remote Job

Yes

Telecommute

Yes

Sponsor Visa

Skills

Product Building, AI Systems, Prompt Design, Model Selection, Latency Budgets, Failure Modes, Eval Pipelines, Analytics, SQL, Data Analysis, Customer Interaction, Problem Solving, Systems Thinking, Agent Building, Code Writing, Specification Writing

Industry

Software Development

Description

Sarvam AI is building India's full-stack sovereign AI platform. We train our own foundation models, build speech and vision systems, and ship products on top of them across consumer, enterprise, edge devices, and hardware. Our customers range from large financial institutions to government bodies. We're a small, fast-moving team where every person ships real product to real users. We're looking for product managers who are builders first. People who talk to customers every week, not because it's on their calendar, but because they can't build well without it. People who've written code, shipped side projects, obsessed over copy, and lost sleep over a bug they couldn't let go of. -------------------------------------------------------------------------------- WHO THIS ROLE IS FOR * 3-7 years of product building experience. You've shipped real products that have reached real people at meaningful scale, at strong B2C or B2B companies in India. * You're a product manager, not transitioning into one. This role is not for people moving into PM from consulting, project management, program management, or other adjacent roles. * You've gotten your hands dirty with AI. You've played around with building agents, you have a feel for how context windows work, you've thought about trade-offs like cost vs. latency vs. accuracy. Maybe you understand evals well, maybe you've experimented with tool calling, maybe you've built a small agentic workflow that actually does something useful. You don't need to check every box, but you should have real hands-on familiarity with how AI products work under the hood - not be planning to learn on the job. -------------------------------------------------------------------------------- WHAT YOU'LL DO This is not a role tied to one specific product line. Based on your strengths and interests, you could end up working across any of our surfaces - consumer AI products, enterprise platforms, developer tools, speech and voice systems, or edge deployments. We'll figure out the best fit together. Wherever you land, the work looks like this: * Own a product area end-to-end. That means talking to users, writing specs, defining what success looks like, shipping, and then measuring whether it actually worked. * Get into the details of how AI systems behave in production - prompt design, model selection, latency budgets, failure modes, eval pipelines. You're not handing a PRD to engineering and walking away. You're in the loop on how things actually get built. * Write code, prompts, and copy when that's the fastest path to learning. Sometimes the best way to test an idea is to build it yourself before pulling in the team. * Dig into analytics to find what's working and what's broken. Set up funnels, write queries, instrument events. Make decisions from data, not decks. * Work directly with engineering, design, and leadership. No layers, no hand-offs. * Ship every week. Not plans. Product. -------------------------------------------------------------------------------- SPIKES & ENERGY You should have done 2-3 things in your life that had real impact. Could be a product you built, a community you grew, a company you started, a hard problem you solved. Product or non-product, doesn't matter. We care that you've gone deep on something and have something to show for it. You can articulate problems clearly, get excited about ideas, and bring energy to a conversation. We're not looking for polished decks. We're looking for people who light up when they talk about what they've built. -------------------------------------------------------------------------------- DATA & METRICS You know SQL. You're comfortable in PostHog, Mixpanel, or similar tools. Not at a "I've used the dashboard" level, but at a "I can write queries and set up funnels from scratch" level. You can define success metrics for a product or feature, not just pick vanity numbers. If something breaks or a metric dips, you can identify the 5 things you'd look at to diagnose the problem. You think in terms of leading indicators, not just outcomes. You use data to make decisions and settle debates, not to decorate a narrative you've already decided on. -------------------------------------------------------------------------------- PRODUCT THINKING & PROBLEM SOLVING You start with the customer's problem, not a feature idea. You understand pain points deeply, both the ones customers tell you about and the ones they don't. You talk to customers regularly. You spend time in sales calls, customer engagement and deeply care about how they use the product. You build a mental model of your users that's grounded in reality, not assumptions. You can take a vague problem space and structure it. Break it down, prioritize, figure out what to build first and why. You're comfortable getting into an uncomfortable level of detail, the kind where most people wave their hands and say "we'll figure it out later." You think big about where things should go, but you know how to cut scope and get v1 out. You ship what you say you will, and you work with engineering to make sure delivery actually happens on time. -------------------------------------------------------------------------------- AI NATIVENESS & SYSTEMS THINKING You should have built at least one working AI agent. Not a demo, not a wrapper around an API, but something that actually does a job end-to-end. Bonus if you've used tools like Claude Code, built skills, or set up agentic workflows. You have opinions on when to use agents vs. simpler approaches. You think in systems. When you look at a product, you see the handshakes between components, the failure modes, the dependencies. You can think about the smallest, most abstract way to build something that still solves the whole problem. -------------------------------------------------------------------------------- CULTURE & OWNERSHIP High ownership. You've stayed up late or woken up early to fix something because you couldn't let it stay broken. You don't wait for permission or process. Builder by default. You have side projects, open-source contributions, or products you've shipped outside of work. Building is not just your job, it's how you spend your time. Ambitious and self-aware. You want to do something meaningful, and you're honest about what you're good at and where you need to grow. High agency. When you hit a wall, you find a way around it. You don't escalate problems. You solve them or come with options. Clear communicator. You can take a complex technical concept and explain it simply, to engineers, to leadership, to users. No jargon for jargon's sake. -------------------------------------------------------------------------------- India's AI moment is here. We're building foundational technology that will touch hundreds of millions of people. This is the kind of work that comes around once in a decade. We're not looking for people who want a PM job. We're looking for the best product builders in the country, people who want to do the most important work of their careers. If that's you, come build with us.

Responsibilities

The role involves owning a product area end-to-end, including user interaction, writing specifications, defining success, shipping, and measuring results across various surfaces like consumer AI, enterprise platforms, or developer tools. Responsibilities also include deep engagement with AI system details such as prompt design, model selection, and failure mode analysis, often requiring writing code or prompts to test ideas quickly.