Product Manager, Voice AI

Wealth management in India has long sat at two extremes: full-service private banking for the ultra-wealthy, and self-directed mutual fund apps for everyone else. Dezerv exists because affluent investors, those with serious capital to deploy, deserve a product built specifically for them: one that combines the judgment of a human relationship manager with the efficiency of digital infrastructure.

That belief produced a phygital wealth platform, a PMS that pairs a digital product layer with dedicated RMs, built for investors who want both access and accountability.

Dezerv operates at Series C scale, backed by institutional investors, with a team of several hundred people building across product, technology, and wealth management.

As Product Manager, Voice AI at Dezerv, you will own the end-to-end build of a voice AI agent that engages prospects across inbound and outbound channels.

You will define the architecture, select the stack, design the conversation flows, and ship the agent from the ground up. The role asks for genuine technical curiosity, hands-on tinkering ability, and the judgment to make hard calls on models, evals, and harness design.

You will report to the Group PM owning Prospect Experience and Engagement.

End-to-end voice AI agent. Own the full product: conversation design, prompt engineering, model selection, and harness architecture.

Stack definition. Evaluate and decide the voice AI stack, from speech-to-speech versus speech-to-text pipelines to the right model for each use case.

Eval framework. Design and run the evaluation harness that tells you when the agent is good enough to ship and when it is not.

Prospect engagement flows. Build the outbound scheduling agent that replaces the human-in-the-loop call booking process.

Inbound query resolution. Own the agent layer that answers prospect questions on how the product works, funding, and onboarding.

Vendor-to-in-house migration. Bring the existing reminder-calling capability in-house and extend it upstream into full scheduling.

Built and shipped a conversational or voice AI product. You have a working agent to point to, not just a course certificate.

Designed prompts and evals yourself. You have written, iterated, and scored prompts without waiting for an engineer to do it for you.

Chosen a model stack for a real use case. You have compared models, weighed trade-offs, and made the call on what to deploy.

Operated without a playbook. You have defined the approach in an ambiguous space and shipped something that worked.

Tracked the voice AI landscape actively. You can speak to what changed in the last six months and what it means for product decisions.

Worked at the intersection of product and engineering. You understand enough of the technical layer to hold your own in architecture conversations.

Tinker before you theorize. Build a working prototype to test an assumption; do not write a spec for something you have not tried.

Own the stack decision. Pick the tools, justify the choice, and live with the outcome; no one will hand you a pre-approved vendor list.

Design evals before you ship. Define what good looks like and how you will measure it before the agent goes live.

Move at the pace the market demands. The voice AI space moves fast; the expectation is that you move with it.

Work in the room. The team is on-site in Bengaluru; alignment happens in person.

Keep the customer in the loop. Every design decision traces back to what a prospect actually needs to hear and when.

Cash compensation in the range of ₹15L to ₹35L a year, depending on experience. You report to the Group PM for Prospect Experience and Engagement and own the voice AI charter end to end. The role gives you the first AI PM mandate at Dezerv, with scope to define how the company engages prospects at scale.