Product Manager Agentic AI

Innefu Labs Limited; New Delhi; On-site; INR 4,000,000 - 7,000,000; Full time

Job summary

Innefu Labs Limited seeks a product leader to own an agentic AI product built from zero, shaping vision, strategy, and MVP scope while collaborating with engineering on orchestration layers and memory management.

You will drive evaluation metrics, PRDs, and enterprise engagement, articulating architecture decisions to leadership, sales, and customers for successful AI-driven workflows.

Qualifications

• 8–12 years in product management with ownership of at least one AI/ML product from concept to live release.
• Technical fluency in agent orchestration and planning, tool calling, and memory patterns.
• Knowledge of AI stack: LLM APIs, RAG pipelines, vector databases, knowledge graphs.
• Familiarity with agent frameworks and MCP for tool/agent interoperability.
• Experience writing PRDs/specs for ML-driven features and defining success metrics.
• Ability to partner with engineering on architecture-level tradeoffs.
• Strong stakeholder management and communication; translate technical tradeoffs to business.
• Experience in enterprise B2B SaaS, security/intelligence, fintech, or data-sensitive domains is a plus.
• Bachelor’s degree in CS/Engineering; MBA is a plus but not essential.

Responsibilities

• Define and own the product vision, strategy, and roadmap for an agentic AI product built from zero — problem discovery, architecture direction, MVP scoping, and scale-up.
• Translate ambiguous problems into phased agentic system design: which tasks are agent-driven vs. deterministic, where human-in-the-loop checkpoints belong, and how autonomy expands release over release.
• Partner with engineering on core architecture decisions — agent orchestration and planning layers, tool-calling design, memory/state management, RAG/retrieval pipelines, and multi-agent coordination patterns.
• Own the evaluation framework for agent quality — task success rate, hallucination/error rate, latency, cost-per-task, and safety guardrails.
• Define prompt, tool, and knowledge-source specifications with applied AI/ML engineers; review agent behavior transcripts and failure cases.
• Run structured discovery with enterprise customers and internal stakeholders to identify high-value workflows for autonomous or semi-autonomous automation.
• Own build-vs-buy and framework decisions in collaboration with engineering and stay current on the agentic AI ecosystem.
• Define and track north-star and input metrics for the product — adoption, task completion, time-saved, cost-to-serve.
• Own the end-to-end roadmap prioritization, sprint scoping with engineering, and release planning; write PRDs, user stories, and acceptance criteria for agent-based features.
• Represent the product externally — customer demos, RFP/technical proposal support, and partner/analyst conversations with technical depth.

Job description

Requirements

• 8–12 years in productmanagement, with demonstrable ownership of at least one AI/ML or agentic AIproduct taken from concept/zero to a live, adopted release — not just a featureadd-on to an existing product.
• Working technicalfluency in agentic AI system design: agent orchestration and planning (e.g.,ReAct-style reasoning, task decomposition, multi-agent coordination),tool/function calling, and agent memory patterns.
• Solid working knowledgeof the surrounding AI stack: LLM APIs and fine-tuning/prompting tradeoffs, RAGpipelines, vector databases, knowledge graphs, and evaluation/observabilitytooling for LLM-based systems.
• Familiarity with commonagent frameworks and protocols (e.g., LangGraph, AutoGen, CrewAI-styleorchestration, or equivalent in-house frameworks) and emerging standards suchas MCP (Model Context Protocol) for tool/agent interoperability.
• Track record of writingclear PRDs/specs for ML-driven features, defining success metrics for probabilistic (not just deterministic) systems, and running structuredexperimentation.
• Comfort partneringclosely with engineering and applied AI/ML teams on architecture-leveltradeoffs — you don't need to write production code, but you need to hold yourown in a systems design conversation.
• Strong stakeholdermanagement and communication skills — able to translate deep technicaltradeoffs into business language for leadership, sales, and customers, and viceversa.
• Prior experience inenterprise B2B SaaS, security/intelligence, fintech, or another data-sensitivedomain is a plus, given Innefu Labs' customer base.
• Bachelor's degree inComputer Science, Engineering, or a related technical field; an MBA orequivalent is a plus but not a substitute for hands-on technical depth.

WhatYou Will Own

• Define and own theproduct vision, strategy, and roadmap for an agentic AI product built from zero— problem discovery, architecture direction, MVP scoping, and scale-up.
• Translate ambiguous,open-ended problems into a phased agentic system design: which tasks areagent-driven vs. deterministic, where human-in-the-loop checkpoints belong, andhow autonomy expands release over release.
• Partner with engineeringon core architecture decisions — agent orchestration and planning layers,tool-calling/function-calling design, memory and state management, RAG andretrieval pipelines, and multi-agent coordination patterns.
• Own the evaluationframework for agent quality — task success rate, hallucination/error rate,latency, cost-per-task, and safety guardrails — and use it to driveprioritization, not just measure it after the fact.
• Define prompt, tool, andknowledge-source specifications working directly with applied AI/ML engineers;review agent behavior transcripts and failure cases as part of the regularproduct cycle.
• Run structured discoverywith enterprise customers and internal stakeholders to identify high-valueworkflows for autonomous or semi-autonomous automation.
• Own the build-vs-buy andframework decisions in collaboration with engineering (agent frameworks, vectordatabases, LLM providers, orchestration layers) and stay current on thefast-moving agentic AI ecosystem.
• Define and tracknorth-star and input metrics for the product — adoption, task completion,time-saved, cost-to-serve — and report outcomes to leadership.
• Own the end-to-endroadmap prioritization, sprint-level scoping with engineering, and releaseplanning; write clear PRDs, user stories, and acceptance criteria foragent-based features.
• Represent the productexternally — customer demos, RFP/technical proposal support, andpartner/analyst conversations — with the technical depth to answerarchitecture-level questions directly.