Product Manager (Gurugram)

Product Manager (Gurugram)

17 Nov
|
Delhivery
|
Gurugram

17 Nov

Delhivery

Gurugram

Role: Product Manager / Senior Product Manager – Agentic AI (TMS)

Location: Gurgaon

What is TMS?

A transportation management system (TMS) is a logistics platform that helps businesses plan, execute, and optimize the physical movement of goods, both inbound and outbound. It provides visibility into day-to-day transportation operations, ensures regulatory and trade compliance along with the associated documentation, and tracks the timely delivery of freight. A TMS streamlines shipping and makes transportation operations easier to manage and optimize.

Why Agentic TMS?

Traditional TMS platforms digitize workflows but still depend on people to run them. Procurement teams run RFQs and chase spot quotes, planners build loads, coordinators chase carriers, analysts audit invoices, and support teams field customer queries. We are building AI agents that take on these jobs end to end:

- Agents that procure freight capacity through spot quotes, and transporter bids
- Agents that plan and optimize loads
- Agents that coordinate with transporter and shippers
- Agents that track and manage exceptions
- Agents that audit and settle freight invoices
- Agents that answer customer queries

These agents run as autonomous systems, with humans in the loop only when needed. They are not copilots bolted onto screens. We are seeking an AI-first Product Builder to join our team in Gurgaon. This is a product management role for someone who builds, not just specs. You will take ideas from customer conversation to working agent yourself, using AI-assisted development to close the gap between “what if” and “it works.” To succeed, you should:
- Enjoy getting your hands dirty
- Thrive on end-to-end ownership
- Let data, not opinion, drive product decisions
- Be excited about defining how AI agents do real operational work in logistics

Expectations from a PM:

- Owns discovery and build of agentic features. This includes identifying which logistics workflows can be automated by agents, and to what degree of autonomy.
- Builds before writing.



You prototype agents, tools, and workflows hands-on to validate an idea in days, then use the working prototype to align customers, design, and engineering.
- Designs agent behavior end to end:
- Goals, decision logic, and the tools/APIs the agent calls
- Context and data sources
- Guardrails, exception handling, and human-in-the-loop checkpoints
- Defines and owns evaluation frameworks for agents. This covers golden datasets, accuracy and decision-quality benchmarks, regression evals before every release, and monitoring for drift in production. Evals are written and run in code, not tracked in spreadsheets.
- Develops and ships production code via AI-assisted tools (Kiro), and submits all changes as PRs. Owns agent prompts, tool definitions, and configs in the repo like any other code.
- Conducts customer discovery, starts with high-level specs, and translates insights directly into product. Everything is documented before productionizing, including agent specs, prompts, tool contracts, and eval criteria.
- Balances agent quality against cost and latency, including model selection, token economics, and the right mix of deterministic logic and LLM reasoning.
- Owns product analytics, workflow QA, release notes, and pilot customers.
- Drives every decision with data, from what to build to whether it worked:
- Sets success metrics and baselines before building, not after shipping
- Analyzes agent decisions and failure patterns to decide what to fix, automate further, or roll back
- Backs product calls with evidence in reviews, and changes course when the data says so
- Owns feature correctness, rollout quality,



customer adoption, and feature-level metrics in production. Agent metrics include:
- Automation/touchless rate
- Decision accuracy
- Escalation rate
- Time and cost saved per transaction
- Builds user trust in agents through explainability, audit trails, and clear handoffs between agent and operator.
- Escalates architectural questions to dev before building, and treats review feedback as blocking.

What we are looking for:

- Undergraduate degree in engineering from a Tier-I institute
- 2+ years of product management experience in enterprise-grade SaaS products or a supply chain organization
- A builder’s track record: products, agents, or tools you have built and shipped yourself using AI tools, ideally including LLM-powered or agentic features in production. Links to things you’ve built are welcome.
- Comfort working in code, including:
- Reading and writing Python or TypeScript with AI assistance
- Working with Git, APIs, and SQL
- Debugging your own work before handing it off
- Strong data orientation: comfortable with numbers and analysis tools, able to work with messy operational data, and has a track record of product decisions driven by quantitative evidence
- Working understanding of:
- LLM capabilities and limitations (hallucination, non-determinism, context limits)
- Prompt and context engineering
- Tool/function calling and agent orchestration patterns
- Experience designing evals or quality frameworks for AI systems, with comfort judging when an agent is “positive enough” to ship
- Exposure to logistics or transportation workflows (procurement, planning, execution, freight audit and payment) is a strong plus
- Strong communication and organization skills, and the ability to think clearly, problem-solve, multi-task, and prioritize
- Ability to thrive in a fast-paced, deadline-driven environment, with the tenacity to develop ideas independently and the conviction to drive them to completion

📌 Product Manager (Gurugram)
🏢 Delhivery
📍 Gurugram

Reply to this offer

Impress this employer describing Your skills and abilities, fill out the form below and leave Your personal touch in the presentation letter.

Subscribe to this job alert:

Get the latest job offers by email for: product manager (gurugram) / gurugram