Meet Joon Lee

Meet Joon Lee, Founding Forward Deployed Engineer of actAVA.ai

Joon Lee is a founding Forward Deployed Engineer at actAVA. Prior to this role, he was a software engineering lead for the Salesforce Agentforce platform. He led AI pilot initiatives within AI Research, translating research-grade AI into real-world enterprise deployments across healthcare, financial services, and consumer technology. Prior to Salesforce, he was an early engineer at Fitbit (later acquired by Google) and a technical consultant to enterprise customers, including large U.S. health systems. With deep expertise in agentic systems and agent orchestration, he operates at the intersection of advanced engineering and customer results. He holds a B.A. in Applied Mathematics and Computer Science from UC Berkeley and brings a mission-driven, results-focused approach to developing and deploying AI agents for our customers.

We took this Friday to interview Joon about the role of a Forward-Deployed Engineer at actAVA and how he helps his customers use agentic AI.

What does a Forward Deployed Engineer at actAVA do?

As an FDE, my role is to ensure that customers get real results and value from our platform—not just a cool demo, but something they actually rely on. That can include everything from understanding the customer's domain and workflow to crafting the system prompt, implementing MCP tools, and creating evaluation metrics. I'm there to do the heavy lifting — connecting our AI to their internal systems, their data, their business logic — so it actually works in context, not just in theory. It's a unique role that combines the technical skills of a software engineer with the client-facing awareness of a consultant. But I feel the most important aspect of this role is earning the customer’s trust at every step of the process. Our customers must be confident that our agents and platform can deliver results safely and securely.

How do you help your customers get over the hard part of agent building?

Getting a working prototype up and running is the first milestone, but earning a customer's trust is what takes it from a pilot to something they depend on. My approach is to start with a simple proof of concept, then iterate, refine, and expand. But in between, there's a lot of work building evals and guardrails—deterministic checks that let the customer verify the agent's output before acting on it. For example, if an agent makes a recommendation that affects a business decision, the customer needs to be able to see the reasoning and verify the math. Once we've crossed the "zero-to-one" hurdle and they can see the agent consistently getting it right, it's much easier to expand its capabilities and let it take on more.

What is your best advice for someone wanting to get into your line of work?

It helps a lot to have a foundational knowledge of backend software — that's your base. But what really sets you apart is staying curious about how people actually work. Anytime I hear about someone's workflow, I can't help but wonder how the current AI tech stack would measure up against their existing process. That instinct — always looking for the gap between what's possible and what's actually happening — is what drives this work. Keep learning the latest AI trends, but pair that with real curiosity about the customer's world. And don't be afraid to fail. That's how you find what works.

Bonus Round: What is your prediction for the biggest AI trend impacting your customers in 2026?

As LLM reasoning continues to improve, customers will spend less time dictating the exact steps of a workflow. Instead, we'll focus on evaluating the results—defining what "good" looks like and letting the agent figure out how to get there. The shift is from programming every step to setting the bar and trusting the agent to clear it. 2026 is the year AI agents have to prove their ROI. The experimentation phase is over. Customers aren't asking "can AI do this?" anymore — they're asking "show me the numbers." That's where evals and observability become essential, and it's where a lot of my work lives.

Thanks for your time today, Joon, and for helping us understand how your customers use AI agents to supercharge their apps. 

Visit us at https://actava.ai/ to learn more.

Meet Joon Lee: View LinkedIn Profile