Product Manager / Senior Product Manager - AI

About the Role

Delhivery moves millions of shipments a day across one of the most complex physical networks: pickups, line-haul, sorting, last-mile, returns, and cross-border. Every one of those steps generates data, decisions, and money on the table. We believe the next decade of logistics will be won by whoever applies AI most intelligently to the physical world. We are building a team of Product Managers who will reimagine logistics with an Agentic first approach. You won't be bolting a chatbot on a dashboard. You will own problems where a better model or a smarter agent changes unit economics, customer experience, or network reliability at national scale. .What You’ll Do

• Find the right problems. Spend time in hubs, with riders, ops leaders, and merchants. Translate messy operational pain into opportunities for creating a seamless flow that scales.
• Own outcomes, not outputs. Define success metrics tied to cost per shipment, delivery success, NPS, or revenue, and be accountable for moving them.
• Ship AI products that real operators use. Design human-in-the-loop workflows, fallbacks, and feedback loops so models earn trust on the ground and keep improving after launch.
• Drive adoption across the org. Work with Operations, Sales, Finance, and Customer Experience to embed AI into how teams actually work.
• Partner deeply with data science and engineering. Shape problem formulation, data requirements, evaluation methodology, and model-versus-rules trade-offs. Understand precision/recall, drift, latency, and cost well enough to challenge and be challenged.
• Run rigorous experimentation. A/B tests, shadow deployments, and phased rollouts across geographies and hubs, with clear go/no-go criteria.
• Stay at the frontier. Evaluate new LLM, agentic, and vision capabilities, and decide pragmatically what to build, buy, or fine-tune.

The kind of problems you could own

• Network intelligence: demand forecasting, capacity planning, dynamic routing, and ETA prediction across the network
• Last-mile optimization: address intelligence and geocoding, delivery-attempt prediction, route and slot optimization, rider productivity
• Computer vision on the ground: shipment damage detection, automated weighing and dimensioning, sort-center throughput, proof-of-delivery verification
• Generative AI and agents: autonomous customer support, ops copilots for hub managers, document and invoice automation, internal knowledge assistants
• Risk and trust: fraud, RTO (return-to-origin) prediction, NDR resolution, and merchant risk scoring
• Pricing and commercial intelligence: dynamic pricing, serviceability, and yield optimization

You will be matched to problem statements based on impact, your strengths, and the stage of each problem. Expect to work across more than one.

What We’re Looking For

Experience

• 2–5 years in product management, ideally with at least 1–2 years shipping ML or AI-driven products (or a strong data science, ML engineering, or analytics background that you have transitioned into product)
• A track record of taking a product from ambiguous problem to measurable business impact
• Strong academic grounding; engineering degree from a top-tier institution preferred, MBA a plus but not required

Skills

• Working fluency in ML concepts: model evaluation, training data quality, feedback loops, and the practical differences between classical ML, deep learning, and LLM-based systems
• Hands-on comfort with data: SQL is a must, Python or notebooks a plus. You pull your own numbers.
• Experience with, or sharp point of view on, GenAI product design: prompt and eval design, RAG, agents, guardrails, and cost/latency trade-offs
• Excellent structured thinking and the ability to communicate with engineers, data scientists, and frontline ops leaders equally well
• Bias to action, ownership, and comfort operating in complex, high-volume, real-world environments

Pedigree signals we value

• Prior experience at product-led or AI-forward companies (marketplaces, e-commerce, mobility, fintech, SaaS, or AI-native startups)
• Evidence of impact: shipped models or AI features with quantifiable results, side projects, or published work
• Curiosity about the physical world: supply chains, operations, and hardware-meets-software problems