Amit Garg and the Expensive Art of Being Correct

Amit Garg wanted the machine to see around corners. At Yale School of Management, he and a physician named Neel Butala imagined using machine learning to predict who might become ill, who might return to hospital and who needed special attention. Their plan was good enough to place among the top five in a university entrepreneurship competition. Then the future arrived early and made an objection: the data was a mess.

The records feeding their elegant idea were scattered across incompatible systems, riddled with gaps and recorded in assorted formats. Garg put the discovery plainly: no matter how good the algorithm, inaccurate inputs would not produce accurate results. The pair could have treated this as an irritating obstacle between them and the interesting work. Instead, they treated the obstacle as the work.

That reversal became HiLabs, founded in 2014 and headquartered in Bethesda, Maryland. Its business is the unglamorous plumbing beneath healthcare intelligence: ingesting, cleaning, reconciling and enriching information so health plans can actually use it. Artificial intelligence attracts attention upstairs. Garg chose the basement, where the pipes have inconvenient names and everything important depends on them.

An engineer learns to read the room

Garg was born in Dehradun, a city in the Himalayan foothills of northern India. He studied industrial engineering at IIT Roorkee, graduating in 2000, and began his career at Infosys. He later joined a startup headquartered in Washington, D.C., where his remit crossed business and information technology. The combination mattered. Code has rules. Organizations have habits, anxieties, incentives and the occasional meeting that feels longer than recorded history.

A 2009 state Medicaid project supplied a defining lesson. Garg was helping replace a claims-processing system that had been in place for 25 years. A rejected deliverable delayed the project and, by his account, cost his employer $2.5 million each month. Management had approved the work. Operations staff, who would have to live with it, had not.

He organized a workshop to break the deadlock. The first day was dreadful. State experts understood Medicaid policy but had little technical experience. Garg's programmers understood the new system but struggled with the state's business processes. Different cultural and communication styles supplied a further layer of static. Everybody possessed a portion of the answer and an excellent reason to distrust the other portions.

The following day, Garg and the operations lead changed the geometry of the room. They divided one large group into smaller teams, mixing people from both sides and balancing expertise, temperament and experience. During the day, he watched the teams. In the evenings, he brought his observations back to his own staff so they could prepare for the next round.

The workshop lasted more than a month. The operations staff eventually approved the system, which went live and, Garg later said, processed more than $20 billion in healthcare expenses. The episode left him with a durable phrase: “thinking beyond you.” The important distinction was not between client and vendor, or winner and loser. It was between a system that worked and one that remained stranded in an argument.

“I strongly believe in focusing on what is right, not who is right.”Amit Garg

His later career carried the lesson into larger systems. At CNSI, he worked on web-based claims processing for state Medicaid agencies and the Centers for Medicare & Medicaid Services. At Northrop Grumman, he led implementation of CMS's Encounter Data Processing System, used in calculating risk-adjusted Medicare Advantage payments. He could see how much policy and money moved through data that was incomplete, delayed or simply wrong.

He went to Yale for the knowledge his engineering education had not been designed to provide: finance, strategy, hiring, organizational culture and competition. In a Healthcare Ventures class, he met Butala, an MD/MBA student fluent in both clinical work and data science. Garg had technological depth but not clinical expertise. Their differences made the partnership useful.

The glamorous future has a filing problem

The founders' first idea had narrative sparkle. Predictive healthcare promises drama, foresight and a machine doing something visibly clever. Data quality is harder to explain at dinner. Even Butala has joked that population health is sexy while data infrastructure sounds boring. Yet a doctor listed at the wrong address, a contract trapped in an unstructured document or a clinical record that arrives in the wrong format can turn administrative untidiness into a real barrier.

HiLabs started with insurers because the economic problem was already present. Provider directories help people decide whether a plan includes the doctors they want. Updating those directories was costly, slow and manual. A company that could automate correction did not have to sell a distant technological dream. It could sell a measurable reduction in work and errors.

By early 2023, Garg told Yale that HiLabs worked with three of the top ten U.S. health plans and several regional plans, covering more than 65 million lives. The company said its MCheck platform had analyzed more than 35 billion records and automatically detected and corrected over three million erroneous ones. Those figures describe scale, but the more revealing measurement may be speed of belief: a decade after the pivot, investors were prepared to finance the premise that cleanliness was a growth market.

In March 2024, HiLabs announced a $39 million Series B. Denali Growth Partners and Eight Roads Ventures co-led the round, with F-Prime Capital participating. The money was directed toward technology, recruitment and product expansion. HiLabs said its provider-directory product was analyzing information on more than 80 percent of U.S. healthcare providers.

Garg's argument about AI remained notably unseduced by AI. He has advised technologists to focus on impact rather than “coolness,” to release and learn instead of waiting for perfection, and to ask why a feature matters before adding it. His caution is equally plain: people should not rely blindly on technology in critical decisions. In his formulation, software increases human efficiency. It does not repeal human responsibility.

“Technology can never replace humans; it just enables us to be exponentially more efficient and productive.”Amit Garg

A company made from corrections

Garg names growth mindset, transparency and perseverance as traits central to his work. Such words can arrive wearing corporate lanyards. His examples give them some abrasion. He describes sharing failures as well as successes inside HiLabs, working alongside customers until systems produce tangible outcomes, and treating iteration as ordinary practice rather than an embarrassing repair.

The company's origin is itself the clearest exhibit. Garg and Butala did not merely modify a feature after customer feedback. They replaced the premise. Prediction gave way to preparation. The interesting machine would have to wait while they fixed the fuel.

There is also a geographical loop in the story. Garg arrived in the United States in the early 2000s and later became a naturalized citizen. HiLabs grew with headquarters in Maryland and research-and-development centers in India. In 2025, IIT Roorkee named him a Distinguished Alumnus for Entrepreneurial Excellence. He has since returned to the campus through HiLabs hackathons, workshops and its sponsorship of the Cognizance technology festival. The student who left Dehradun to study engineering came back asking students to build healthcare intelligence agents.

His stated horizon extends beyond provider, claims and clinical records. Garg has mentioned genomic, pharmacy and clinical-trial data as possible areas for the same infrastructure. The ambition has not shrunk since Yale. It has acquired a sequence. First make the information accurate. Then make it move. Only then ask it to predict.

That sequence also explains why Garg talks about customers as collaborators. Healthcare systems do not arrive as blank sheets. They contain old investments, entrenched workflows and people whose practical knowledge rarely fits neatly into a product specification. Garg has argued that disruption can come through coexistence, with new tools designed to fit alongside what organizations already use. It is an unfashionably diplomatic definition of innovation, but it rhymes with the Medicaid workshop. The goal is not to make the incumbent system lose an argument. The goal is to get useful change adopted. At HiLabs, that means combining the pattern-finding reach of software with business rules, operational context and the judgment of subject-matter experts. The machine can expose contradictions across many records. A person still has to understand what a correction will do once it leaves the screen and enters somebody's working day.

In an industry fond of announcing revolutions, Garg's career offers a slyer proposition. Progress may begin with correction. The decisive founder is sometimes the one willing to say the original idea cannot survive contact with the facts. HiLabs exists because two people wanted to see the future and discovered, profitably, that the present needed cleaning first.