Surya Nelakanti hadn't written a single line of code. And the feature shipped anyway. That sentence would read as a punchline at most companies. At Abnormal AI, it's a Tuesday. One year into his role as a Software Engineer on the Human Security Products team, Surya's experience building AI Phishing Coach tells a specific story about what engineering looks like when a company is designed from the ground up to put AI in every engineer's hands.

The Signal That Said "Go"

Before Abnormal, Surya was a full stack engineer at a drone mapping startup. When GPT 3.5 dropped, his priorities shifted. "It became obvious to me that I had to move to a company that was investing heavily in utilizing LLMs to enhance dev productivity," he said.

The interview confirmed it. "My AI-enabled interview was the signal that made me confident that Abnormal was the right company to be in for the next phase of my career." He wasn't looking for a company that talked about AI. He was looking for one that had already changed how it worked because of it.

Ownership From the Start

Within his first few weeks, Surya was handed a core feature of AI Phishing Coach: building the system that generates phishing simulation templates from real customer threats. The feature required cross-functional collaboration across teams and time zones, a far cry from the small core team at his previous company.

"It felt overwhelming, but good, to have the level of trust placed in me from the beginning itself," he said. "I love hitting the ground running. It felt like my skills were being utilized in the way that I like."

The intimidation faded once he started doing the work. "In the process of doing, you realize that a lot of the unknowns slowly disappear," he said. "It also allows you to better refine your plans as well." That pattern, building your way through the ambiguity, became the operating rhythm for everything that followed.

The Feature He Didn't Code

The story that captures Surya's experience most precisely is the Executive Dashboard. The feature gives CISOs and security leaders a real-time view of their organization's phishing resilience. And Surya built it with a workflow that would be unrecognizable to most engineers.

"I'd point the AI at the codebase and let it explore, then make the architectural call, then review what it produced," he explained. When the AI's first attempt built a filter dropdown instead of a generic response structure, Surya caught it and redirected. The AI rebuilt the entire approach in minutes. When testing revealed a bug where training completion exceeded 100% due to a cross-window counting edge case, he described the issue, the AI researched root causes and proposed fixes. Each cycle of plan, direct, review, iterate took minutes instead of the hour-plus it would take to write and debug manually.

Today, he's trending toward all of his tasks being at least 90% completed by AI.

What It Means for Engineering

Surya sees this as more than a productivity hack. He sees it as a fundamental shift in what engineering is.

"A big part of software engineering has always been reusability. We build on frameworks we didn't create. We implement algorithms that were defined decades ago," he said. "AI takes that principle to an extreme. Syntax, boilerplate, the translation of intent into code all become a solved problem."

The cost of trying something drops to near zero. "You stop over-planning in your head and start trying things, because the cost of being wrong is nearly zero." And what's left, he said, is the work that always mattered most: understanding the problem, making the right design calls, and knowing the system well enough to catch what the AI can't.

For product engineers specifically, Surya sees the role blurring with product management. "That boundary between 'what to build' and 'how to build it' used to be heavily demarcated. AI is collapsing it, and engineers who lean into that ambiguity will thrive."

A Company That Hands You the Keys

What makes this possible isn't Surya's individual curiosity. It's the infrastructure Abnormal provides.

"If you have a strong enough reason to use a new AI tool, Abnormal is ready to get you a key or a license to utilize it," he said. The company doesn't restrict the number of tokens engineers can use, which gives Surya room to explore models at the cutting edge, compare them against each other, and figure out the right workflow for each type of task. He uses Nora and Glean for internal documentation, ChatGPT for researching and ideating, and Wispr Flow for speech-to-text.

"I am constantly using the latest and greatest models, comparing all of them with each other, and figuring out the right workflow and models to use to better create features," he said.

Building Things People Love

The work connects to something beyond speed. When Surya's PM returned from a customer call with feedback about the quality of the phishing simulation templates, the impact landed differently than a Jira ticket moving to "Done."

"As someone who got into software engineering because I liked creating websites for people," Surya said, "knowing that customers love what we are building is something that resonates."

And the product itself sits at the center of a real problem. "The people writing real phishing emails are already using AI," he explained. "If your security awareness training is still using generic, static templates, you're training people for yesterday's threats. AI Phishing Coach flips that equation: it generates hyper-personalized simulations that match the sophistication of what's actually hitting inboxes."

A Year In, a Career Reshaped

His team uses Bonusly to celebrate wins big and small, and quarterly team dinners mark the milestones. But the deeper shift for Surya is how the past year has changed his understanding of what his career can be.

"By being allowed to use cutting-edge models and being incentivized to actually use AI in the right way, it has fundamentally changed my perception of software engineering as a career," he said. The engineer's job is moving upstream. Less time translating intent into syntax. More time deciding what to build and why.

And for anyone interviewing? "The level of ownership you get, even for a company at this stage, is pretty similar to an earlier stage startup."

That combination, startup-level ownership with the tooling, resources, and mission of a company protecting thousands of organizations worldwide, is what made Surya's first year look nothing like a typical new grad experience. It's what made it look like the future.

Abnormal AI is redefining what it means to build as an engineer. Explore our open roles and join us.