There is a particular kind of expertise that only comes from reading hundreds of other people's homework. Sasibhushan Rao Chanthati, who goes by Sasi, has that expertise. He is a Senior Software Engineer at Hirekeyz, and for two years running, 2024 and 2025, he has been recognized as a BIG All Star Judge for Information Technology, on top of judging duties for Globee and Stevie.
Sasi sat down with Russ Fordyce for the Winners' Circle podcast to talk about what all that judging has taught him, and honestly, the conversation goes a lot further than awards. Sasi also happens to be a hands on engineer building real systems inside a major financial firm, a published researcher with four technical books and more than ten peer reviewed papers, and the creator of an original AI system for detecting something almost nobody in tech leadership wants to talk about out loud, IT burnout.
A judges eye view of what actually separates a strong nomination
Ask Sasi what draws him to judging and he does not talk about titles or prestige. He talks about pattern recognition. After scoring nomination after nomination, he has learned that the strongest ones never lean on company size or revenue. They lean on capability, meaning the specific thing a team built and how it actually gets used inside the organization. Weak nominations describe the mission. Strong ones describe the mechanism.
That distinction matters more than ever because of what Sasi is seeing across the board, agentic AI moving from buzzword to baseline. Gartner predicts that 40 percent of enterprise applications will feature task specific AI agents by the end of 2026, up from less than 5 percent just a year earlier, and Sasi's own work with ServiceNow gives him a front row seat to that shift.
AI cost management just became universal
One trend Sasi keeps circling back to is FinOps, the discipline of managing cloud and now AI spending. According to the FinOps Foundation's State of FinOps 2026 report, the share of organizations actively managing AI costs jumped from 31 percent in 2024 to 98 percent in 2026. That is not gradual growth. That is a discipline going from niche to mandatory in the span of two years.
Sasi has watched this play out from inside the room. Companies that once built their own financial operations tooling now almost universally buy software as a service instead, funneling everything through one platform rather than maintaining a patchwork. It mirrors a broader shift he has watched throughout his career, the move from on premise legacy systems to cloud infrastructure on AWS, Azure, and Google Cloud, which picked up dramatically around the pandemic and never really slowed down.
Small models, big shift
Here is something that might surprise people outside the engineering trenches. The biggest, most expensive frontier language models are not always what enterprises actually deploy. Sasi describes a pattern where companies increasingly favor smaller, fine tuned models trained specifically on their own data, rather than routing every query through a massive general purpose system. It costs less, it is more secure, and frankly, a model trained on your organization's actual workflows does not need to know the entire history of the internet to answer questions about your incident queue.
Security remains the deciding factor here too. Sasi is blunt about it: cheaper AI products from less established vendors often cut corners on the layered security that regulated industries require, and in his world, that tradeoff is rarely worth the savings.
The problem nobody wants to admit: IT burnout
The most personal part of the conversation is Sasi's own research into IT burnout, work he began building back in 2020 and 2021 after securing copyright protection through the US Copyright Office. His system analyzes patterns in workplace communication, chats, emails, and messages tied to a unique employee identifier, using vector embeddings and a large language model to flag signs of stress before they become a resignation letter.
What makes the project compelling is how it evolved. Sasi's first version required manual review. His second used vector embedding and semantic search to automate analysis. The current iteration measures results continuously and repeats the process automatically, surfacing alerts to HR when someone shows signs of being overwhelmed. According to CIO.com's coverage of the ongoing burnout crisis in technology, the industry has struggled with an epidemic of overwork for years, and Sasi's approach represents one of the few attempts to actually measure it systematically rather than treat it as an inevitable cost of doing business.
It is a timely question given where the industry is heading. Gartner predicts that investment in generative AI will drive a 20 to 30 percent reduction in customer service and support agents by 2026, a shift that will ripple through the very teams Sasi's burnout research is designed to protect.
Where this leaves engineers and IT leaders
Sasi's closing thought during the conversation is worth sitting with. AI is supposed to make people more productive, but productivity gains have a habit of expanding to fill the available hours rather than shrinking the workday. The engineers building these systems are often the same people most exposed to the burnout risk they create. Sasi's answer is not to slow down AI adoption. It is to be deliberate about measuring the human cost alongside the efficiency gains, something his own research has spent years trying to make possible.
Congratulations to Sasibhushan Rao Chanthati on being named a two time BIG All Star Judge for Information Technology. You can see his full judge profile and hear the complete conversation, including his take on shadow AI risk in regulated industries, on the Winners' Circle podcast.
Subscribe to the Winners' Circle podcast to catch every conversation with this year's honorees.