MoMo
AI tóm tắt
Skills bắt buộc
This role owns the quality bar for Moni. You define what a good answer looks like for an AI assistant serving Vietnamese merchants/users, build the system that measures it, and give the final quality sign-off before release, so quality becomes an evidence-based decision instead of a matter of opinion. You are accountable not only for the quality numbers, but for the merchant and business results those numbers are meant to produce, as Moni becomes a leading AI assistant in Vietnamese fintech. Mô tả công việcQuality Standards Evaluation System: Own the definition of quality for Moni — the failure categories, the metrics and the thresholds themselves. Design and maintain the evaluation system: gold-set test datasets, automated scoring, and human review of sampled conversations, kept valid as the assistant expands to new surfaces and more complex, multi-step tasks. Give the final quality sign-off before each release, and be accountable for what reaches merchants. Failure Analysis Improvement Loop: Build the process the team uses to sort and categorise real conversation failures, not only the analysis you run yourself. Separate one-off errors from repeated, systemic ones. Decide where each fix belongs — prompt, knowledge base, retrieval, intent design, hand-off logic, or model choice — and make the trade-off between quality, cost and response speed explicit. Own an improvement plan with a committed quality gain each cycle. Knowledge Intent Coverage: Set up how the knowledge base is owned and kept accurate across domain owners: who owns what, how often it is reviewed, one approved source for each answer, and compliance sign-off where needed. Decide which intents to build next based on how often merchants ask, how much it matters to them, and how costly a wrong answer would be — so the assistant covers more real questions without inventing answers. Business Impact for Merchants Users: Own how quality metrics connect to business results (AI adoption, task completion, deflection, retention). Agree targets with leadership, work with the data team on tracking and dashboards, and explain each quarter what the numbers mean and what should change — including recommending that work be stopped when the numbers do not move. Customer Insight Pain Point Discovery: Run merchant interviews, surveys, field visits and log mining on a regular schedule to find needs the assistant does not serve yet. Turn them into product opportunities with a clear hypothesis, an estimated size, and a success measure. Cross-functional Delivery: Work with AI engineering, design and operations to write requirements that cover edge cases, drive sprint planning and UAT, remove blockers, and ship on schedule with quality accepted before release. Represent the quality position when the roadmap requires a trade-off, including with senior stakeholdeYêu cầu công việcJob Requirements — Knowledge / Skills
Analytical mindset: can read hundreds of raw conversations and build a clear, non-overlapping set of failure categories that other people can apply consistently — and explain how the conversations were selected so the results are representative.
Data-driven decision making:Comfortable writing your own queries and working with dashboards, funnels and cohort data. Knows when a change in the numbers is real and when it is noise. Defines the metric before proposing the solution.
Hands-on experience with an AI assistant or LLM product: has shipped or meaningfully improved one. Understands prompting, knowledge base retrieval (RAG), how to evaluate answer quality, and how hand-off to a human agent should work — and can reason about the trade-off between quality, cost and speed. Machine learning or research background is nice-to-have required.
Product execution depth: has owned a product area end-to-end — user flows, requirement specs, edge-case thinking, release readiness — with an engineering team working in sprints.
Judgment in a regulated business: understands why a financial assistant must be accurate, sourced and compliant, and works with legal and risk rather than around them.
Clear communication: can explain a technical quality problem to a business stakeholder, a business goal to an engineer, and can defend a decision to hold a release using evidence.
Quality-obsessed: sets the standard rather than meeting it; reads real merchant conversations regularly and is uncomfortable with "good enough" answers.
Fast learner in a moving field: the AI landscape changes every few months; actively follows what is new, tests it, and brings what works into the product instead of waiting to be taught.
Customer-centricity: understanding and genuinely caring about small business owners or end-users.
Evidence over opinion: brings data to disagreements, changes their mind when the data says so, and still makes the call when the data is incomplete.
Ownership in ambiguity: builds the process where none exists, and leaves it running for others.
Bachelor's degree in Business, Economics, Engineering or a related field.
3+ years in product management, product operations, business analysis, including demonstrated hands-on work on an AI assistant, LLM product or B2C product.
Has owned a product metric end-to-end — from defining it and setting up tracking through to a measurable improvement.
Experience delivering in sprints, including running UAT and accepting quality before release.
Plus: fintech, payments or B2C products experience; exposure to the Vietnamese SME or merchant segment; familiarity with AI tools.
Nguồn: careers MoMo
