A camera is a peculiar colleague. It never blinks, never asks for coffee and remembers everything badly until someone teaches it what matters. For most of its history, the camera’s job was passive: collect the scene, preserve the evidence, wait for a human to arrive. Krishna Khadloya built his career during the long change in that job description. The machines around him learned to notice a person, a plate, a parcel, a pedestrian or an unusual sound - and to turn recognition into an event another system could use.
Khadloya is president and CEO of IntelliVision, the San Jose company that develops video and audio analytics for security, transportation, access control, driver monitoring and smart cameras. His title puts him in the current AI boom. His résumé places him somewhere more interesting: in the infrastructure that preceded the boom, when performance had to be measured, mobile software had to be managed and video intelligence had to fit inside the unglamorous constraints of devices, networks and customer installations.
The route begins with electrical and electronics engineering at BITS Pilani in India, followed by a master’s degree in computer science at SUNY Albany. Later came an executive program at Stanford’s Graduate School of Business. The sequence reads like a compact map of his working life: hardware, software, then organizations. Before 1999 he led research and development at Mentor Graphics, in electronic-design automation. Then he joined Keynote Systems, whose business was measuring how the web and mobile services actually performed for users.
Chapter oneThe apprenticeship in reliability
Khadloya stayed at Keynote for more than a decade. He joined in September 1999, served as director and senior director of engineering, and became vice president of engineering in April 2006. By 2011, he remained in that role. It is the kind of tenure that resists the modern habit of treating a career as a brisk tour of logos. He had time to see systems age, teams grow and yesterday’s clever shortcut become tomorrow’s support ticket.
Web-performance software has an unforgiving premise. Customers do not buy the explanation for why a service felt slow; they buy a reliable account of what happened. Measurement has to travel through real networks, across locations and devices, while preserving enough context to be useful. Those disciplines reappear later in Khadloya’s work. Video AI also begins with a noisy world. It must decide quickly, operate at scale and communicate a result that another person or machine can trust.
After Keynote, Khadloya served as vice president of engineering at MobileIron, an enterprise mobile-security company. The device had moved from desk to pocket; the management problem followed. At IntelliVision he would move closer still, toward cameras and other edge devices. Across the changes, the recurring question was where intelligence should live. In a central system? In the cloud? Near the sensor? The useful answer is usually an architecture, not a slogan.
“We are excited and looking forward to the new chapter and journey for IntelliVision.”Krishna Khadloya, November 2024Chapter two
When pixels acquire verbs
At IntelliVision, Khadloya led engineering and products before becoming CEO. The company’s vocabulary is full of nouns that want to become verbs: face, plate, person, vehicle, sound. A frame contains a license plate; a system detects it, isolates it, reads it and assigns confidence. A camera contains a person; software determines presence, motion or identity. A microphone contains an environment; a model classifies an event. The commercial value appears only after perception becomes action.
Khadloya’s name appears with collaborators on patents that make that progression concrete. The subjects include scalable cloud video, security systems with face recognition, human-presence detection at the edge, pedestrian detection for driver assistance, license-plate recognition, audio-event classification and enrollment systems that can update reference information over time. More recent grants extend the pattern to railcar identifiers and other transportation uses.
A patent list can become an ornamental wall of jargon. Read sideways, though, and this one tells a clearer story. The camera is no longer only an eye. It is a small operating post, deciding which information deserves bandwidth and attention. Edge processing matters because sending every raw frame elsewhere is slow and expensive. Confidence matters because a weak match should not be treated like a fact. Continuous enrollment matters because the world, and the people in it, refuse to stay visually convenient.
The company changes hands
In November 2024, IntelliVision entered a new corporate chapter. Nipun Vision Corporation acquired full ownership from Nice North America and said IntelliVision would again operate as a standalone entity. Khadloya took the president and CEO job. Vaidhi Nathan, IntelliVision’s founder and former CEO, returned as an investor and chairman. The structure paired an experienced internal operator with the company’s founding memory.
Khadloya described the brief in two directions. Existing products and customers needed continuity. The newly independent company also intended to pursue the broader AI market and emerging generative-AI technologies. Every mature technology business knows this tension: maintain the machine that pays today while redesigning the one that must pay tomorrow. The challenge is particularly sharp in physical security and transportation, where installations live longer than fashion cycles and reliability cannot be patched with enthusiasm.
One career, four computing eras
Electronic-design software at Mentor Graphics: building tools for people who build systems.
Web and mobile performance at Keynote Systems: measuring experience across unruly networks.
Enterprise mobility at MobileIron, followed by engineering and product leadership at IntelliVision.
Standalone IntelliVision under Nipun Vision: edge-to-cloud analytics with a wider AI mandate.
His preparation for that assignment is less theatrical than cumulative. Electrical engineering supplies respect for the device. Computer science supplies the abstraction. Performance monitoring supplies suspicion of ideal conditions. Mobile infrastructure supplies an appreciation for distributed control. Years of leading teams supply the final correction: a system diagram is not an organization chart, however much executives may wish otherwise.
IntelliVision says its products work across security and safety, access control, driver and occupant monitoring, intelligent transportation and smart-building markets. These categories sound separate until one notices their common appetite: each wants a machine to watch continuously and interrupt a person only when something deserves attention. The product is partly recognition and partly restraint. A system that alerts on everything has invented a more expensive form of silence.
The names beside Khadloya’s on the invention record matter here. Vaidhi Nathan appears on work involving cloud services, human presence, audio and virtual assistants. Chandan Gope is a collaborator on pedestrian and audio systems. Manuel Gonzalez joins him on multi-factor recognition and enrollment methods. Ilya Popov and other IntelliVision engineers appear on license-plate and railcar recognition. The pattern is collective, as serious engineering usually is. One person may help set the direction, but a product crosses the finish line through people who understand different failure modes.
That detail also sharpens the word “leader.” Khadloya’s official biography emphasizes globally distributed teams. Distributed engineering is often described as a calendar problem: find the tolerable meeting hour, rotate the pain, record the call. The harder version is architectural. Work has to be divided so that teams can move without continually surprising one another. Interfaces must be explicit. Ownership must be legible. A decision made in one place needs enough context to survive its trip to another. The same virtues make edge-to-cloud systems work.
There is a quietly comic symmetry in leading software that watches for exceptions while managing an organization designed to prevent them. Both systems need thresholds. Both need escalation paths. Both become useless when every small variation triggers an alarm. A good executive, like a useful analytics model, cannot spend the day shouting that something moved. Judgment begins with deciding which movement changes the story.
The operator’s wagerTeach the machine what deserves a second look
Generative AI adds another layer to Khadloya’s long-running question. Detection asks what is present. Generation and multimodal reasoning can help describe, search and connect events. Yet the old engineering demands remain impolitely alive: latency, compute budgets, privacy boundaries, false alarms, device compatibility and the plain fact that customers already own equipment. The future arrives through an installed base.
That may be why Khadloya’s career feels less like a sudden AI conversion than a long approach. Mentor Graphics, Keynote, MobileIron and IntelliVision each placed him near a boundary between a complicated system and the person depending on it. The interfaces changed. The obligation did not. Make the hidden work legible. Make the result reliable. Build a team that can keep doing both after the launch slides are archived.
The camera, meanwhile, keeps acquiring verbs. It detects, recognizes, classifies, counts, searches and alerts. Under Khadloya, IntelliVision’s next task is to decide which new verbs are useful enough to ship and disciplined enough to trust. He has called it a new chapter and journey. The phrase is optimistic, but the career beneath it is practical. Before a machine can notice what matters, an organization has to make thousands of smaller judgments correctly. Someone must teach the camera. Someone must also teach the company.