He watched a bad hire at his last company derail a quarter. He didn't fire the person. He rebuilt how companies find people - from the data layer up.
Hariharan "Hari" Kolam • Findem, Redwood City
When Hari Kolam was scaling Instart Logic internationally - the CDN and web delivery company he co-founded as CTO - his team moved fast. Too fast. A couple of hires didn't work out. Quarters slipped. Good people spent months cleaning up after the wrong ones.
Most founders absorb this as the cost of growth. Kolam absorbed it as a data problem. "The cost of a bad hire has a cascading effect and is extremely hard to undo." That sentence became the founding logic of Findem.
He didn't look at the HR industry and see bureaucracy in need of process improvement. He looked at it and saw a 1940s data format - the resume - still powering 21st-century talent decisions for trillion-dollar companies. The resume: a flat, static, self-reported document invented before the transistor.
Findem launched in 2019-2020 on a single contrarian premise: the real problem isn't candidate sourcing. It's candidate data. Fix the data and the sourcing, matching, diversity, retention - all of it - gets dramatically better.
"People search remains unchanged since World War Two... now digital footprints are spread everywhere. We wanted to essentially expand the scope of information so thereby you can have an educated decision."
"Resumes are neither complete nor consistent. They don't capture the true impact a professional has made. At Findem, we saw an opportunity to redefine the data set and provide talent teams with richer, verified information through 3D profiles."
"Talent decisions aren't an AI problem - they're a data and business intelligence problem."
- Hariharan Kolam, CEO, FindemKolam earned his B.S. in Computer Science from Visvesvaraya National Institute of Technology in India, then crossed the Atlantic for a master's at Stony Brook University in New York. Two continents, two CS degrees, and a clear trajectory toward systems engineering at scale.
His first industry stop: Sun Microsystems, where he contributed to the Solaris Cluster group - the fault-tolerant, high-availability operating system layer that kept enterprise infrastructure running. From there, he moved to Akamai and then Aster Data, where he worked across the entire development stack from kernel-level code to BI application layers. The breadth was unusual. Most engineers specialize. Kolam went wide on purpose.
When he co-founded Instart Logic, he brought that full-stack thinking to CDN technology - web performance, content delivery, dynamic page acceleration. The company accumulated over 50 patents with Kolam as a named co-inventor, a body of intellectual property that spans distributed systems, content optimization, and network architecture.
The pivot from web delivery to HR technology looked lateral from the outside. From the inside it was the same problem: massive, messy, distributed data that needed to be structured, indexed, and made queryable in real time. The substrate was different. The engineering instinct was identical.
Findem's core product is what Kolam calls "3D profiles" - candidate records assembled not from resumes but from over 100,000 external sources: GitHub commits, patent filings, publications, company databases, professional networks, news mentions, and more. The result is a profile with over 10,000 searchable attributes per candidate, versus the approximately 20 facts in a typical resume.
The platform processes 1.6 trillion data points and applies machine learning to generate "Success Signals" - attributes that distinguish candidates who genuinely thrive in specific roles. It's the difference between searching for "product marketing manager who worked at Google" and accidentally surfacing an agency intern who managed a Google advertising account.
Kolam frames AI's role in recruiting with a sharp distinction: "There's an IQ side and an EQ side to recruiting. The IQ side - building pipelines and generating candidate slates - can be largely automated with AI. The EQ side, like meaningful candidate conversations, is where recruiters will continue to shine."
Findem's acquisitions of Glider AI (assessment and interview intelligence) and Getro (relationship-driven talent networks) are extending this logic end-to-end - from first-touch sourcing through screening to offer.
Expanding Findem's expert-labeled dataset and accelerating domain-specific AI - teaching machines to understand talent decisions the way exceptional hiring managers do, with full context and nuance.
Kolam watched the AI hype cycle hit HR tech from an unusual vantage point: a CEO who had already co-authored 50+ technical patents and built data infrastructure at scale before the generative AI wave arrived. His response wasn't enthusiasm or skepticism. It was precision.
"AI becomes performative instead of transformative when underlying data is broken or unreliable." The line is short but it explains most of what is wrong with AI-first HR tools. Large language models trained on the open web inherit all of the internet's noise, bias, and incompleteness. A resume written to pass an ATS screen is not training data. It is signal-corrupted noise.
Findem's answer is an expert-labeled dataset - human recruiters and hiring managers annotating outcomes so the model learns what "success in this role" actually looks like. Then Success Signals: patterns extracted from those verified outcomes that predict future performance.
"Generic AI can write job descriptions or summarize resumes, but it can't understand contextual hiring nuances," Kolam says. His bet is that domain-specific AI - built on structured, verified, expert-labeled talent data - outperforms general-purpose AI for the single most consequential decision companies make: who to hire.
"AI can't automate judgment it doesn't understand - capturing recruiter and manager expertise is essential."
"By elevating talent data from a flat commodity into a rich strategic asset, we're the only company making it AI-ready."
"Domain-specific AI, built on expert-labeled data and Success Signals, raises that ceiling considerably."
Kolam built diversity into Findem's data model, not as a filter layered on top but as a structural property of how candidates are discovered. Traditional keyword search is structurally blind to diverse talent pools - it recycles the same networks, the same schools, the same job titles.
Findem's attribute-based search exposes candidates who fit a role by demonstrated capability rather than pedigree. The platform includes compliant prioritization for underrepresented candidates with full visibility into diversity metrics across every stage of the hiring funnel.
Kolam's view: diverse teams aren't just more equitable - they measurably outperform. "Diverse teams significantly outperform competitors in profitability, innovation, and retention." This is his argument for why DEIB belongs at the data layer, not the policy layer.
Kolam reads biographies. Not business frameworks - biographies of people who built things that mattered and then had to rebuild them when the world changed. He follows NPR's "How I Built This" with the same intensity. IBM's comeback story from "Who Said Elephants Can't Dance" is a reference he returns to: the willingness to abandon a dominant position, rethink the core product, and bet everything on a new model.
His father was an early mentor - a formative influence on how Kolam thinks about responsibility, persistence, and the long game. The phrase "you only fail when you stop trying" reads less like a fortune cookie and more like a lived operating principle when you've watched him build two companies from scratch across two continents.
He's not loud about the mission. He's specific. When Kolam says building exceptional teams is the single most important factor in a business's success, he says it as an engineer who once miscalculated and paid the price - not as a pundit making a PowerPoint point.
"Building exceptional teams is the single most important factor in a business's success."
- Hari KolamThe $51M Series C isn't Kolam's exit - it's his expansion budget. The money goes toward building out the expert-labeled dataset that makes Findem's AI genuinely domain-specific: human expertise encoded as machine-readable Success Signals at a scale no competitor can easily replicate.
His stated ambition: transform the entire HR function from task automation to genuine transformation. "Together, we're turning static talent data into a living, strategic engine that not only fills roles but predicts and shapes the future of work."
The Getro and Glider AI acquisitions show the shape of that vision - a single platform from market intelligence to role definition, candidate discovery, assessment, and offer. The resume, the spreadsheet, the twelve-tab ATS workflow - all of it collapsing into one verified, data-rich system where AI handles the cognitive load and humans handle the human part.