Do you want to be a cog in the machine or build the machine?
That’s the question that VP of Data and AITimothy Wongposes to job applicants considering a career atAirwallex.
What Does Airwallex Do?
Airwallexis an AI-native financial operating system helping modern businesses manage payments, global accounts, spend management, and embedded finance.
Wong believes that Airwallex offers technologists a unique opportunity to define an industry. Unlike large tech companies and AI labs, Airwallex gives its people “total architectural freedom on fintech’s hardest AI problems” — bolstered by an $11-billion valuation, $1.3 billion in annual recurring revenue and 74 percent year-over-year growth.
“You get agency and global scale, not one at the expense of the other,” Wong said.
On Wong’s teams, every workday offers a new chance to shape the global financial landscape.Whether they’re developing new AI agents to expedite the decision-making process or helping build a robust knowledge layer to effectively manage structured and unstructured data globally, data and AI employees pioneer innovative solutions — and advance their careers in the process.
“My teams in particular are using AI and the latest technology to drive real business value across product, sales and marketing, simultaneously, at global scale, in a regulated environment,” Wong said. “That combination is genuinely rare.”
Below, Wong shares more about the unique work he’s tackling at Airwallex, the types of individuals who thrive on his teams, and the greatest benefits of building a career at the company.
Timothy Wong
VP of Data and AI • Airwallex
What It’s Like to Work in Data and AI at Airwallex
Airwallex is building the AI that runs global finance. We’re designing autonomous systems that make real decisions in payments and compliance for more than 676,000 businesses worldwide.
My teams in particular are using AI and the latest technology to drive real business value across product, sales and marketing, simultaneously, at global scale, in a regulated environment. That combination is genuinely rare. Most companies get one or two of those. We’re navigating all of them at once, which is what makes the problems hard and the work interesting.
What Do Airwallex’s Data and AI Teams Do?
Airwallex’s data and AI teams build knowledge platforms, intelligence products and AI capabilities that power decision-making, customer experiences and growth across the company. This team develops customer-facing products such asKai AI assistant, drives internal AI transformation and builds AI powered GTM systems that drive revenue growth.
What Airwallex’s Data and AI Teams Own
The easiest way to understand the org is to start with Knowledge Platform, because everything else builds on top of it.
Historically, teams like ours worked with structured data: feeding machine learning models and running pipelines. With AI, the requirements have changed completely. AI can now read docs, markdown files and all kinds of unstructured data. But the classic failure mode is garbage in, garbage out. If you don’t invest enough in the knowledge layer underneath, none of the applications work well. Knowledge Platform is the team solving that foundational problem: How do you synthesize, index and serve both structured and unstructured data globally, keep it fresh and make it compliant across multiple jurisdictions? The industry hasn’t fully solved this yet, which is part of what makes it an interesting place to work.
AI engineeringbuilds the application layer on top of that foundation, primarilyKai,our AI assistant, and the internal framework we use to orchestrate agents across our financial products. The core challenge there is that we’re asking probabilistic AI systems to operate in a deterministic financial environment. That means a lot of careful work on guardrails, evaluation and system design: What must be deterministic, what can be non-deterministic, and how do you make those boundaries safe?
Data sciencehas evolved a lot. We’ve merged data scientists and data engineers into one team of full-stack data builders. The old model of someone who just runs analyses doesn’t really hold anymore. Everyone on the team is expected to build ML models, AI agents and evaluation frameworks. We have data scientists working on payment optimization and growth optimization. We have people building what we internally call AirDS. They’re essentially AI agents that can interact with the context a senior leader needs to make better decisions faster.
Growth and go-to-market (GTM) engineeringis the team using AI to reimagine how growth actually works. That covers three areas: product-led growth (personalized recommendations across our product surface, driving activation and retention); sales-led growth (we call it Revenue OS: rebuilding the sales workflow from scratch using AI, so reps can close deals faster and account managers can spot the right interventions at the right time); and marketing-led growth (SEO, SEM and lifecycle marketing with AI agents). We’re also building an autonomous sales agent product, where the revenue operations team defines the ideal customer profile, and AI handles the prospecting, research and outreach. Everything is bundled under one product called AirGTM, with the central vision of driving GTM velocity through personalized, insight-driven action and effortless collaboration.
We’re the only team under the CTO that explicitly owns both technological and business outcomes. That intersection is uncommon. If you like working close to where technology decisions become business decisions and having those decisions actually matter, this is the right place.
What you build gets seen by senior leadership. We have regular product reviews with our CRO and CEO, while major technology and platform decisions are reviewed directly with our CTO.
The Hardest AI Problems Airwallex Is Solving
We’re working on frontier problems that the industry is still figuring out. If we were just implementing things that are already solved, that wouldn’t be worth highlighting.
The first is evaluating AI analytics without ground truth. There’s no standardized way to assess whether an AI-generated analysis is good or bad, and in financial decision-making, the cost of being wrong is very high. We have people dedicated to researching this, exchanging ideas with peers who are thinking about it seriously, and it’s genuinely open territory.
“We’re working on frontier problems that the industry is still figuring out.”
The second is orchestrating agents in a regulated, globally distributed environment. We have dozens of product domains, multiple APIs and compliance requirements that vary by region. Getting AI agents to orchestrate across that correctly, with proper evaluation end to end, while staying compliant — that’s a hard system design problem, not a toy problem.
The third is model strategy. We’re moving past “one model fits all.” We’re actively figuring out how to route specific tasks to different models, how to fine-tune open-weight models on our proprietary data, and how to balance latency, accuracy and cost. This is something the whole industry is working through right now. We’re working through it with real financial data and real stakes.
AI Products and Systems Airwallex Is Building
Kai, our AI financial assistant, is live in the product today. Customers interact in plain language and Kai takes action.AgentOSis the internal framework we built to govern and orchestrate agents across Airwallex’s financial products, so an agent can reconcile transactions or pay invoices within a live account. We’ve also positioned ourselves for where finance is going. We recently launchedAiri, our agentic wallet. No one has fully defined what an agentic wallet means yet, but we’re already building for that future.
We’re applying the same ambition inside Airwallex. Our AI transformation isn’t about giving everyone a chatbot. We’ve mapped hundreds of workflows across the company and are redesigning how work gets done around humans and AI. We’re building shared knowledge and context infrastructure, an internal AI platform that lets teams create and automate their own workflows, AI-native analytics where agents increasingly participate in analysis and decision-making, and AI-native GTM systems that help our commercial teams research, prioritize and act. The goal is to move from AI assisting individual tasks to AI becoming part of how the company operates.
None of this is a someday story.
What AI Products Is Airwallex Building?
Airwallex is developing products and systems including Kai, its AI financial assistant; AgentOS, its framework for governing and orchestrating AI agents; and Airi, its agentic wallet.
How Airwallex Gives Technologists Autonomy and Ownership
We hire builders. Even at my VP level, I have been very hands-on. I want everyone on the team to have genuine end-to-end ownership.
The scope differs by level, not the expectation of ownership. Staff and above on the platform side will be driving foundational architectural decisions, making the right bets over the next 12 to 18 months so we don’t get vendor-locked and we keep flexibility as the technology evolves. Those decisions get reviewed in detail with our CTO.
On the application side, you own the full stack. You’re building directly with business stakeholders, shipping prototypes fast, showing something in a day, iterating and showing it again. We deliberately don’t run long two-week sprints before anyone sees anything. The feedback loop is tight by design.
What Qualities Help People Succeed on Airwallex’s Data and AI Teams?
According to Wong, those who perform well on his teams are curious, use AI daily and are willing to get their hands dirty with business problems, not just implementation. “The technology is still immature, similar to where machine learning was 15 years ago,” Wong said. “People who thrive here can let go of the “this is how we’ve always done it” instinct and rebuild their understanding from scratch with engineering rigor.”
Wong shared that the ones who tend to struggle on his teams are those who rely on past playbooks. “We need people who can break technology down from first principles and make sound decisions with incomplete information,” Wong said.
At an AI lab, you’re closer to frontier models but further from the full production system. At mature platforms like Big Tech, the foundational architecture is already locked in. You have to ask yourself, do you want to be a cog in the machine, or build the machine? And while startups offer speed, they often lack a real data surface and customer base.
At Airwallex, you get total architectural freedom on fintech’s hardest AI problems, plus the upside of a rapidly scaling, pre-IPO company. We have a valuation of roughly $11 billion, $1.3 billion in ARR, and 80 percent year-over-year growth. You get agency and global scale, not one at the expense of the other.
What Career Growth Looks Like on Airwallex’s Data and AI Teams
Honestly, with AI, we see less need for pure people managers going forward. What we do need are people who can take on increasingly complex, high-leverage problems. The paths are architecture-heavy or application-heavy.Foundational architecture is deep technology, with less emphasis on direct business outcomes. Application work is more balanced, as you’re driving toward product and business KPIs.
What we don’t offer, and aren’t planning to build, is a track for people who want to eventually exit building entirely and just manage. If someone’s goal is to stop doing hands-on work, this probably isn’t the right fit.
Why Tech Talent Chooses Airwallex in Singapore and the U.S.
We operate with incredibly high talent density across both hubs.
Singapore is a genuine hub for us, not a satellite. We’re seeing high-caliber people with founder backgrounds joining as engineering leads, and that shapes the culture. Singapore is also where we’re deeply engaged on AI policy and regulation, which matters for a company operating at Airwallex’s scale and license footprint. The domain knowledge embedded in the Singapore team is real.
The U.S. presence, particularly for the growth function, is part of a broader Americas expansion. Building a team of data scientists and engineers in the United States isn’t arbitrary; it’s because our primary commercial focus there requires people embedded in that market. The teams operate in pods, with a strong emphasis on knowledge-sharing globally, but with enough independence that the time-zone friction doesn’t kill productivity.
How Airwallex Shows Its AI Strategy Goes Beyond Hype
If you genuinely don’t believe AI is a meaningful industry shift, you shouldn’t join an AI team — mine or anyone else’s. That belief has to be real.
For Airwallex specifically, look at what’s actually in production. Kai is live, making real decisions for customers. Airi positions us well in the agentic commerce space, not a company scrambling to retrofit AI onto legacy infrastructure. Our CEO has been publicly and specifically bullish on AI, not as a marketing position, but as an operational bet the company is actively funding.
The product releases speak for themselves. That’s the cleanest answer to “Is this hype?”
What do the data and AI teams at Airwallex do?
The data and AI teams at Airwallex build knowledge platforms, intelligence products and AI capabilities to power decision-making, customer experiences and revenue growth. Team members develop customer-facing tools like the Kai AI assistant while driving internal AI-native transformation across workflows. They directly own both technological and business outcomes simultaneously across product, sales, and marketing at a global scale.
What AI products is Airwallex developing?
The company’s products include Kai, an AI financial assistant that allows customers to interact in plain language and execute actions within accounts. Other solutions include AgentOS, an internal governing framework designed to orchestrate and manage AI agents across financial products (e.g., reconciling transactions or paying invoices), as well as Airi, an agentic wallet built for the future of agentic commerce.
What is the Airwallex Knowledge Platform?
The Airwallex Knowledge Platform is the foundational data layer upon which all other Airwallex AI applications are built. It’s designed to synthesize, index and serve both structured and unstructured data (such as docs and markdown files) globally. The platform ensures data stays fresh and remains compliant across multiple international regulatory jurisdictions to prevent “garbage in, garbage out” failure modes in downstream AI models.
Airwallex is Hiring|View All Jobs
