Written by Kathryn Zhou
When I joined Decagon as an Agent Engineer, I didn’t really know what to expect. A year in, it’s turned out to be one of the most rewarding and technically challenging roles I could’ve imagined.
Building at the frontier
Agent engineering is technically challenging because we’re constantly building at the frontier of what AI agents can accomplish. Beyond understanding how to wire together agent prompts, knowledge bases, tools, and different channels of communication, I’ve had to dive into the nitty gritty of how we can build and deploy these agents at scale. I’ve written scrappy code to benchmark time-to-first-token in an effort to squeeze out seconds of latency in production. I’ve built robust reliability infrastructure, from retries and failover mechanisms to monitoring and recovery systems, to keep customer agents consistently available even when upstream dependencies are unreliable. And I’ve spun up my own experiments across different prompts and workflows to build and test agentic reasoning.
Shaping the product
Stepping back from the code, I also get to see why the work matters: improving how companies build agents on Decagon and how end users experience those agents.
As we take on large customers with increasingly bespoke requirements, we don’t accept building one-off point solutions as an answer. Instead, we treat these requirements as an opportunity to improve the core Decagon product itself. For example, when our team partnered with Block, the parent company of Square and Cash App, we needed to deliver faster, more natural voice interactions that genuinely felt human, alongside a broader set of product improvements. We challenged ourselves by asking, How can we deliver these features to the platform so that everyone, not just Block, can use them? How do we build flexibility while still providing structure and guidance? Those are the kinds of questions that make being an ASWE especially exciting at Decagon. Agent engineering sits at the center of product strategy, which means we have meaningful ownership and influence not just over implementation, but over product direction, platform design, and customer outcomes.
Owning customer outcomes
This customer work drives the future of our platform and ultimately our business too. Agent Engineers directly impact deals and revenue, and there is nothing more exciting than getting that assurance that our partners loved the work we put in and want to collaborate for years to come.
With all of the hats I get to wear, the role really becomes choose your own adventure. Personally, I’ve found a lot of joy in interfacing directly with customers and winning deals, but there are many different paths within Agent Engineering. Some people lean more heavily into deep technical architecture, others into product engineering or customer strategy. I love that people are able to carve out their own style and space while still being part of a team. Growth in the role comes from taking on larger and more complex problems – whether that’s leading bigger customer engagements, owning larger product scope, or helping scale the team itself.
Building together
The best part of being an ASWE is getting to collaborate across many different roles at Decagon. The energy in the office is infectious: people whiteboarding ideas out in the open, huddling around someone’s desk debugging an issue, pulling each other into discussions across teams. Agent engineers work extremely closely with orchestration, research, infrastructure, product, and APMs, and that cross-functional collaboration creates a very fast-moving and intellectually exciting environment. That energy was one of the main reasons I was sold on joining Decagon in the first place.
If you love building quickly, staying close to users, and pushing the limits of AI agents, Agent Engineering is one of the most impactful places to be right now. And if this sounds exciting to you, we'd love to talk.
.png)
