The AI Enablement team at Zalopay applies Large Language Models (LLMs) and Natural Language Processing (NLP) to automate and optimise internal operation workflows - reducing manual effort, accelerating processing times, and improving output quality across the business.
As an Associate Product Owner, you will receive well-scoped automation problems from your line manager, translate them into clear product requirements, and rigorously evaluate whether the LLM-powered solutions actually deliver against those requirements. The role is focused on output quality and workflow automation - not UX or user-facing product design.
You will collaborate closely with
AI Engineers, Customer Success, and domain teams, sitting at the translation layer between operational requirements and technical implementation. We run disciplined sprint cycles using Jira and Confluence, with daily stand-ups, backlog grooming, sprint planning, and outcome tracking.
What You Will Do
Write Product Requirements
Receive problem statements and automation opportunities from your line manager, then translate them into structured product documentation - PRDs, user stories, and acceptance criteria
Break down complex operational workflows into clear, testable requirements that AI Engineers and domain teams can build and validate against
Maintain requirements documentation in Confluence and track progress in Jira across sprint cycles
Evaluate LLM Outputs
Design evaluation frameworks to assess whether LLM-powered automation meets the defined acceptance criteria - covering accuracy, consistency, edge case handling, and failure modes
Build test sets, define scoring metrics, run structured evaluations, and document findings clearly
Identify gaps between expected and actual model behaviour, and work with AI Engineers to iterate - adjusting prompts, refining criteria, or flagging data quality issues
Track evaluation results across iterations to measure progress and validate that improvements hold across varied inputs
Collaborate Across Teams
Work closely with AI Engineers on prompt design, model constraints, and output quality - translating operational requirements into technically actionable specs
Coordinate with Customer Success, Operations, and other domain teams to gather ground truth labels, validate evaluation criteria, and ensure automated outputs meet operational standards
Participate in sprint ceremonies - stand-ups, grooming, planning, and reviews - and keep documentation up to date
Requirements
Must Have
Proficiency in Python.
Graduated with a degree in Computer Science, Data Science, or Applied Mathematics, with solid CS fundamentals: algorithms, data structures, and databases is a strong plus but not required - you'll be expected to pick these up quickly on the job if you don't already have them
Strong structured thinking: able to decompose a workflow problem, write unambiguous requirements, and reason carefully through edge cases and failure modes
Detail-oriented - small specification errors have downstream consequences in automated processes; you take precision seriously
High initiative and comfort with ambiguity - you will often work in areas without a clear answer and will need to figure things out systematically
Intellectually curious and think beyond - stay curious, constantly ask why certain things work that way and always think of how can I make it better
Nice to Have
Hands-on experience with NLP libraries or LLM tooling - Hugging Face, LangChain, LlamaIndex, OpenAI API, or similar
Familiarity with prompt engineering techniques or LLM evaluation methodologies