Must Haves: 

• 6+ years in product ownership, technical program management, or engineering on a large-scale transactional commerce platform
• 3+ years owning a platform or domain end to end.
• Hands-on SAP Commerce Cloud (Hybris) experience or similar cart engine (if not, then open to any ecommerce experience is fine) 
• Azure DevOps at working depth — work item hierarchy, WIQL queries, boards, wiki, and pipelines, GitHub 
• SAFe PI planning experience.
• Experience with LLMs/AI
• Fluency within context engineering, dev skills (ability to read Java and Spring configuration) 
• Excellent organization, verbal communication, written communication
• Ability to manage multiple workstream

Plusses: 

• Grocery, retail, or omnichannel commerce domain experience — especially EBT SNAP/WIC, or another regulated tender or payment-compliance surface.
• Order management and fulfillment integration experience (OMS, marketplace/3P, store pickup and delivery).
• Working knowledge of Kafka event-driven integration, Redis caching and distributed locking, and Azure SQL performance characteristics.
• Observability practice — OpenTelemetry instrumentation strategy, span and attribute naming conventions, sampling policy, and dashboard and alert design. This platform has an active OTel and Observe migration in flight.
• Major platform upgrade or migration experience (JDK or Spring major version, or a commerce platform replatform).

Day to Day:

Client's SAP Commerce Cloud (Hybris) platform is the cart and order engine behind every digital order Client takes — online grocery, general merchandise, third-party marketplace, and store pickup and delivery. It processes 25,000–36,000 orders a day and roughly 12 million OCC API requests every 24 hours, serving the CMS React web storefront, the Client Mobile App, and Backoffice customer-service agents.

We are looking for a Technical Product Owner who can own this platform at the level of detail it actually demands. This is not a role that translates business requirements into tickets and hands them to engineers. The Hybris platform is a 43-extension Java 21 codebase on SAP CCv2 with deep integration into payments, EBT/SNAP/WIC tender, tax, inventory, order management, and marketplace systems. Its failure modes are database saturation, thread exhaustion, cron drift, and concurrency defects — and its highest-value work is diagnosed from production telemetry, not gathered from stakeholders.

The person in this role reads production logs and traces, writes the root-cause analysis, sizes the fix with engineering, sequences it against a SAFe PI backlog of over 1,000 items, defends the tradeoff to architecture and business leadership, and signs off on the release that carries it. You will be the single accountable owner for what the platform does, what it stops doing, and what it costs to run.

What You Will Own

• The cart-to-order engine. Cart and cart-group (multicart, including 1P and third-party Mirakl marketplace carts), checkout and place-order, and the full order lifecycle — placement, edit, cancel, replace, status. Hybris is currently the system of record for digital orders, with a long-term migration path to Yantra OMS that you will help shape.
• Regulated tender. EBT SNAP and EBT WIC multi-tender checkout, WIC benefit balances and product restrictions, and mixed-tender orchestration across Aurus, Visa, Mastercard, AMEX, Discover, Maestro, Client Credit Card, and EBT. This logic spans hundreds of Java files across roughly nine extensions and carries real compliance exposure. It is the most sensitive surface on the platform.
• Promotions and rewards pricing via LightningCart, one of the platform's largest data domains and a recurring source of both customer value and production risk.
• Subscriptions, tip orders (Pre-TIP and TIPS), and Simply Give donations.
• Platform health and cost. Capacity, latency, error budget, and SAP Commerce Cloud consumption. Azure SQL vCore headroom, memory entitlement, and API replica sizing are product decisions with a dollar figure attached, and they belong to this role.
• Integration contracts. OCC API versioning and backward compatibility for AEM, MMA, and Backoffice consumers; Kafka feeds for product, price, stock, ATP, store, and third-party data; order export through Order Broker to OMS; and the Okta / Azure APIM identity path.
• Deliberate retirement. Special Buy retirement, ALOD order-export transport wind-down, legacy REST v1/v2 sunset, and Vertex removal. Knowing what to stop running is as much of this job as knowing what to build.

Core Responsibilities

Own the backlog and the roadmap

• Own a Portfolio Epic → Capability → Feature → Story/Enabler → Bug hierarchy in Azure DevOps across a backlog exceeding 1,000 active items, and keep it ordered by defensible reasoning rather than requester volume.
• Run PI planning and sprint commitment on a SAFe cadence (PI/sprint releases such as 52.3 and 53.1, with Innovation & Planning sprints), coordinating across the platform and shopping teams.
• Balance a genuinely hard portfolio: net-new customer capability, performance and stability enabler work, regulated-tender compliance, technical-debt paydown, and vendor-driven upgrades. Make the tradeoffs explicit and write them down.
• Maintain the documentation of record — ADO wiki pages, architecture decision records, and the domain glossary — so decisions survive personnel change.

Diagnose production, not just describe it

• Investigate production incidents directly using SAP Cloud Logging, Dynatrace, Azure SQL Query Store, HAC, and FlexibleSearch. Produce root-cause analysis with named signals, counts, time windows, sources, and explicit confidence labels.
• Distinguish a symptom from a cause. Recent examples from this platform: a 100% Azure SQL DTU saturation event traced to a 17-month-old unindexed polymorphic query; Tomcat thread exhaustion caused by a 5:1 mismatch between the thread pool and the Redis connection pool; three of ten monitored cron jobs silently not executing in production, leaving failed order processes with no retry path; and a single cart containing 3,926 units of one item generating roughly 41,000 database statements per price refresh, degrading order placement site-wide.
• Separate durable fixes from immediate mitigations, and hold the line on scheduling the durable one.
• Report platform health in terms leadership can act on — before-and-after evidence, order-flow impact, SLA position, and cost — not dashboards.

Own release readiness and the release decision

• Run pre-deploy risk review on release branches: flagged changes, blast radius, E2E automation coverage gaps, rollback plan, and mixed-version pod overlap risk during rolling deploys.
• Provide Technical Product Owner sign-off as a formal Go/No-Go gate, coordinating CRQ change records, SAP support cases, and rehearsed rollback with a validated RTO.
• Own post-release verification: matched before/after telemetry windows, error-signature checks, and a written post-release report.

Manage the integration and vendor surface

• Enforce OCC change-impact tracking so consuming teams are never surprised. No OCC field is removed without a versioning strategy.
• Coordinate with Order Broker, Orders Finalization, Orders Pricing, Payments, OMS (Yantra), MOMS, MI9, BOPAS, Mirakl, Instacart, Bazaarvoice, Akamai, and the Kafka data-integration teams.
• Manage SAP as a vendor: support cases, Enhanced Operations reporting, DBA engagement, and commercial consumption against licensed entitlement.
• Partner with Architecture/ARB, Security, SRE, DBA, QA, Observability, FinOps, and Customer Care.

Raise the platform's engineering leverage

• Continue the AI-assisted engineering program already funded on this team: an in-repo agent governance layer, five purpose-built MCP servers (build and test tooling, guardrailed production query, HAC, log analysis, Dynatrace), and eighteen repeatable skills covering release-readiness scanning, deterministic daily order reporting, ImpEx and Groovy change safety, PR summarization, and TDD.
• Push recurring manual analysis toward deterministic, tested automation — the daily production order report was moved from LLM inference to scripted extraction with a twenty-case golden-file test suite, and that is the standard.