High-velocity engineering teams have a tall order: move quickly without sacrificing quality, support large workflows and meet the needs of different departments.
Matt English, an engineering and product manager at marketplace platform Taskrabbit, works on a team that owns Taskrabbit’s IKEA integration end-to-end. The scope forces a certain kind of dynamic, he said: “Engineers who understand the business problem, not just the ticket.”
“The team’s strength is ownership — people dig into root causes instead of shipping a patch and moving on,” English said. “That habit of asking ‘what’s actually happening here’ before ‘how do we fix the symptom’ is what keeps trust high, both inside the team and with the partner.”
On the team at fintech platform Adyen, Head of AI James Sun keeps the AI research team moving forward quickly by focusing on the company’s mission and highlighting the impact they have on customers every day.
“Amid the fast-moving AI landscape, it is easy to get engrossed in the technical details of optimizing models or scaling systems, but one has to periodically zoom out to see the bigger picture and remember why,” Sun said.
Want to know the secret to fast-moving, high-performing engineering teams? We spoke with Sun and English to hear their secrets to success.
Adyen is a financial technology platform that provides end-to-end payments capabilities, data-driven insights and financial products.
What makes your team dynamic work?
The Adyen AI Research Team dynamic thrives on intentional diversity of skill, rigorous ownership and a low-ego culture. We deliberately assemble a multidisciplinary team with complementary skill sets, allowing us to proactively cover blind spots and elevate our collective technical problem-solving.
We cultivate an owner mentality that treats engineering as a specialized craft. Every team member strives to take end-to-end accountability, designing and building products with an uncompromising bar for quality, motivated by a shared drive to produce work we are genuinely proud of.
Underpinning this approach is a deeply collaborative culture. We prioritize low-ego team players who go out of their way to support one another. By fostering a high-trust atmosphere where continuous peer-to-peer learning is paramount, we consistently up-level individual capabilities, the team and cross-functional partners alike.
How do you balance speed and cohesion?
We balance speed and cohesion by pairing individual autonomy with pragmatic alignment around shared goals.
Velocity relies on hiring self-motivated, autonomous talent committed to continuous learning.
Given that AI/ML development relies heavily on experimentation, we empower team members to be bold, propose ingenious solutions and iterate through failure without bureaucratic friction. We remain strictly metrics-driven, using data to validate hypotheses quickly and keep high-velocity experimentation tied to core outcomes.
Cohesion is built on psychological safety, mutual trust and respect. Everyone is encouraged to propose new ideas or challenge the status quo, regardless of title or tenure. This egalitarian approach keeps communication honest, ensures alignment on primary objectives and maintains high cohesion across projects.
What keeps your team grounded?
Our team stays grounded through a commitment to Adyen’s global mission and the real-world impact of our technology. Amid the fast-moving AI landscape, it is easy to get engrossed in the technical details of optimizing models or scaling systems, but one has to periodically zoom out to see the bigger picture and remember why. We maintain perspective by remembering who we serve — merchants and consumers reliant on secure financial systems at scale.
We operate on the belief that rigorous execution and an enjoyable work environment are mutually reinforcing. Creating exceptional work requires a culture that balances intense focus with fun, camaraderie and shared purpose.
Ultimately, the AI research team at Adyen stays anchored by keeping the human aspect front and center. Fostering a collaborative, fun and mission-driven environment ensures our ideas move beyond theoretical experimentation to create meaningful, global impact in service of businesses and shoppers around the world.
Taskrabbit is a marketplace platform that conveniently connects people with Taskers to handle everyday home to-dos, such as furniture assembly, handy work, moving help and more.
What makes your team dynamic work?
Our team owns Taskrabbit’s IKEA integration end-to-end — pricing and the platform work that keeps a massive retail partnership running. That scope forces a certain kind of dynamic: engineers who understand the business problem, not just the ticket. I sit across engineering, product and account management for this partnership, so decisions get made with full context instead of getting lost in handoffs. The team’s strength is ownership — people dig into root causes instead of shipping a patch and moving on. That habit of asking “what’s actually happening here” before “how do we fix the symptom” is what keeps trust high, both inside the team and with the partner.
How do you balance speed and cohesion?
Speed without structure just moves chaos faster, so anything with real complexity gets a short plan before anyone touches code — what we’re doing, in what order and who signs off. For anything with more than one reasonable path, we lay out the tradeoffs rather than quietly picking one. That upfront alignment is what lets us move fast later, because we’re not re-litigating decisions mid-build. We lean on Jira and Confluence so tickets and design docs stay cross-linked — anyone can trace a decision back to its rationale. And we treat corrections as standing rules, not one-off fixes: if something breaks a second time, that’s a signal the process needs to change, not just the code.
What keeps your team grounded?
A shared bug-severity framework — P0 through P4, with clear off-hours expectations — keeps severity and urgency calls objective, so we’re not debating them in the middle of an incident. We also hold a hard line on verifying before shipping: tests pass, logs checked, root cause confirmed, not just “seems fixed.” And when something feels hacky or like it’s papering over a deeper issue, we say so out loud rather than letting it slide. That combination — a clear severity language, a verification habit and permission to call out shortcuts — keeps the team honest with itself even when the pace is high.
