New opportunity

Senior Product Director, AI & Agent (Chatbot)

OKX · Singapore

About this role

The Senior Product Director, AI & Agent (Chatbot) will own the full AI product agenda at OKX, focusing on both user-facing conversational AI and internal agentic systems. This role is responsible for building the AI layer that defines the company's next decade, ensuring products are intelligent, scalable, and auditable. Responsibilities include: User-Facing AI: Leading the strategy and execution of a crypto-native conversational AI assistant for onboarding, trading, and support. Operational AI & Agent Systems: Owning the roadmap for internal agentic systems in compliance, risk, fraud, and operations, ensuring outputs are traceable and regulatorily defensible. AI Platform: Designing shared infrastructure for orchestration, evaluation, safety, and governance. Requirements: 12+ years in product and tech, with at least 5 years in a senior leadership role owning technically complex products at scale. Proven track record shipping conversational or agentic AI products. Deep familiarity with LLM product architecture, including multi-agent systems, RAG, tool use, evaluation, and safety design. Experience operating in regulated environments with strict auditability and explainability requirements. Strong consumer product instinct and executive-level communication skills in English and Mandarin. Preferred qualifications include a background in crypto, fintech, or financial services at a global scale, and a graduate degree in a technical field. Benefits include a competitive total compensation package, L&D programs, wellness and meal allowances, and comprehensive healthcare schemes.

For full details about this role, click “Apply on company site” to view the employer’s original posting.

Skills for this role

Conversational AIAgentic AILLM product architectureMulti-agent systemsRAGTool useAI evaluationSafety designProduct strategyFinancial marketsProduct leadership

Checking your profile…

Preparation suggestions, not the employer’s actual interview questions. Always check the original posting for current requirements.

Understand the role

What matters most

Unified AI Product Strategy
The role requires balancing two distinct mandates: a user-facing conversational assistant for trading and support, and internal agentic systems for compliance and risk. You must demonstrate how to build a shared AI platform infrastructure that supports both, ensuring scalability and governance across the entire OKX ecosystem.
Regulatory Compliance and Auditability
Because OKX operates in a highly regulated financial environment, all AI outputs—especially those in risk and fraud—must be traceable and defensible. You must prioritize the design of safety, evaluation, and governance layers that satisfy strict audit requirements while maintaining product performance.
Technical Leadership in LLM Architecture
You are expected to lead the technical direction for multi-agent systems, RAG, and tool use. This requires deep expertise in evaluating AI performance and ensuring that the underlying architecture is robust enough to handle the complexity of global financial markets.

Likely challenges to think through

Balancing Innovation with Financial Risk
The core challenge is deploying cutting-edge agentic AI in a high-stakes financial environment where errors can lead to significant regulatory or financial consequences. Prepare by articulating a framework for 'human-in-the-loop' systems and rigorous testing protocols that mitigate hallucination and operational risk.
Cross-Functional Stakeholder Alignment
Managing the AI agenda involves working with compliance, operations, and product teams, each with different risk appetites. You will need to demonstrate how you communicate complex technical trade-offs to non-technical stakeholders and align them on a unified roadmap that prioritizes long-term platform stability.

Practice your answers

Technical practice

Designing Multi-Agent Systems for Financial Operations

What to coverExplain how you would architect a multi-agent system to handle complex tasks like fraud detection or compliance. Focus on how you manage agent hand-offs, state persistence, and ensuring that individual agent actions remain traceable and auditable.

RAG Implementation in Regulated Environments

What to coverDiscuss the trade-offs between retrieval accuracy and latency when implementing RAG for a crypto-native assistant. Address how you ensure the retrieved context is both accurate and compliant with financial data privacy standards.

AI Evaluation Frameworks

What to coverDescribe your approach to building an evaluation framework for conversational AI. How do you define metrics for 'trust' and 'accuracy' in a high-stakes trading environment, and how do you automate these evaluations during the CI/CD process?

Ensuring Explainability in Agentic Outputs

What to coverGiven the requirement for regulatorily defensible outputs, how do you design for explainability in LLM-driven decisions? Suggest techniques to bridge the gap between 'black box' model outputs and the transparency required by financial regulators.

Safety Design and Guardrails

What to coverWhat is your strategy for implementing safety guardrails in a user-facing chatbot that handles sensitive trading information? Discuss the balance between preventing harmful outputs and maintaining a helpful, fluid user experience.

Tool Use and API Integration

What to coverHow do you manage the reliability of LLM tool use when agents interact with core trading or wallet infrastructure? Address error handling, validation of tool outputs, and preventing unauthorized or unintended state changes.

Scaling AI Infrastructure

What to coverAs you build shared infrastructure for orchestration and governance, how do you ensure it remains performant as the number of internal and external agents grows? Suggest strategies for modularity and configuration management.

Data Governance for AI

What to coverHow do you approach data governance when training or fine-tuning models on financial data? Focus on the lifecycle of data from ingestion to model inference, ensuring compliance with global financial regulations.

Behavioural practice

Leading Through Ambiguity in AI Strategy

What to coverDescribe a time you defined a product roadmap for a nascent technology where the path forward was unclear. How did you align stakeholders on the vision, and what metrics did you use to track progress in the early stages?

Managing Stakeholders in Regulated Industries

What to coverProvide an example of how you have navigated conflicting requirements between product innovation and strict regulatory or compliance constraints. How did you communicate the trade-offs to executive leadership?

Driving Cross-Functional Execution

What to coverTell me about a time you led a complex, technically demanding project that required collaboration between engineering, legal, and operations teams. How did you ensure all teams remained aligned on the product's goals and timelines?

Building Trust in AI Products

What to coverShare an experience where you had to earn user trust for a new product feature that involved automation or AI. What specific actions did you take to demonstrate reliability and transparency to the end-user?

Sharpen your knowledge

Topics to prepare

Regulated AI Governance and Auditability
The role requires building systems that are regulatorily defensible in fintech. Exercise: Draft a framework for an AI audit trail that captures LLM decision-making steps, inputs, and confidence scores for high-stakes operations like fraud detection or compliance, ensuring explainability for regulators.
Multi-Agent System Architecture
The position involves designing shared infrastructure for agentic systems. Exercise: Map out a high-level architecture for a multi-agent system where specialized agents handle distinct tasks like user intent recognition, trade execution, and risk assessment, focusing on how these agents communicate and maintain state.

Skills to revise

LLM Evaluation Strategy
The role demands robust evaluation of AI outputs. Exercise: Define a set of quantitative and qualitative KPIs to measure the performance of a conversational AI assistant, specifically focusing on accuracy in financial contexts, hallucination mitigation, and user trust metrics.
RAG and Tool Use Integration
You must lead the execution of conversational AI that interacts with live trading data. Exercise: Design a workflow where an LLM uses RAG to retrieve real-time market data and executes a tool-based action, such as checking a user's wallet balance or initiating a trade, while maintaining strict safety guardrails.
Executive Communication for AI Strategy
You will define the company's AI agenda for the next decade. Exercise: Prepare a 5-minute executive summary pitch in both English and Mandarin that articulates the business value of transitioning from manual operational workflows to agentic AI systems, emphasizing ROI and risk management.

Your preparation checklist

Prepare a case study on agentic system auditability
Develop a structured walkthrough of a past project where you implemented AI in a regulated environment. Focus specifically on how you ensured outputs were traceable, explainable, and compliant with strict regulatory standards.
Draft a framework for LLM evaluation and safety
Create a one-page document outlining your approach to evaluating LLM performance and safety in a high-stakes financial context. Include your methodology for RAG accuracy, hallucination mitigation, and guardrails for multi-agent orchestration.
Synthesize a product strategy for crypto-native AI
Prepare a brief proposal on how a conversational AI assistant can improve user onboarding and trading experiences in a crypto exchange. Highlight how you would balance user-facing innovation with the necessity of institutional-grade trust.
Refine executive communication in English and Mandarin
Prepare to discuss complex technical architectures and product roadmaps in both English and Mandarin. Practice articulating the business value of internal agentic systems for compliance and risk operations to non-technical stakeholders.
Formulate strategic questions for the leadership team
Prepare questions regarding OKX's current AI infrastructure, the specific regulatory challenges they prioritize for agentic systems, and how the AI product team integrates with existing engineering and compliance departments.

Availability last checked: 18 Oct 2026. Status can change after a check; confirm on the employer’s website. Report an issue · Our checking policy

KEEP EXPLORING

Similar AI jobs

Related skills and specializations in Singapore. Matched to this role, not your profile.

Explore more jobs

Finding similar opportunities…