Ashish Vengsarkar likes a grid with consequences. In a crossword, one letter can rescue a corner or ruin every answer that crosses it. In an optical network, one design choice can ripple through power budgets, capacity, cost and reliability. His working life has moved between those two forms of construction: the network carrying traffic and the puzzle carrying language. Both reward a patient eye for constraints. Both become interesting when the pieces refuse to fit.
This is not a hobby pasted onto a résumé for color. Vengsarkar began solving crosswords at 10. In college he constructed British-style cryptics, built from wordplay, anagrams and misdirection. In 2005 he shifted to the American grid, where the dense crossings impose another order of difficulty. A year later, his first puzzle appeared in The New York Times. By 2022, he had published 19 there.
While that catalogue grew, so did the scale of his other grids. He worked on fiber at Bell Labs, built optical products in startups, led Nistica through two corporate transactions, joined Google’s global networking organization and returned to the startup floor as chief executive of nEye.ai. His field stayed recognizable even when the customers, traffic and stakes changed. Light remained the medium. The challenge was always how to route it more usefully.
The fiber education
Vengsarkar trained as an electrical engineer at the Indian Institute of Technology Bombay, then completed a doctorate at Virginia Tech in work centered on fiber optics. Bell Labs gave him the setting in which research met infrastructure. He investigated how digital information, encoded as pulses of light, could travel more effectively through optical fiber. His name appears across inventions involving dispersion-compensating fiber, gratings, waveguides and optical filters.
Those technologies sit below the visible internet. A person taps a screen; light carries the consequence through glass. Vengsarkar later offered a more tactile description: the builders of optical networks make and maintain the fiber-optic pipes through which distant datasets meet. It is an engineer’s view of connectivity, stripped of the cloud metaphor. Traffic still needs a path. Every path has physical limits.
Research was only the first layer. He moved into product management and learned to translate laboratory performance into something a carrier could install, operate and buy. He also earned an Executive MBA from Wharton, graduating as a Palmer Scholar. The pairing became durable: scientific depth on one side, markets and organizations on the other.
A failure worth keeping
In January 2000, Vengsarkar co-founded Photuris, an optical-systems startup built for metropolitan networks. The timing was brutal. The telecom boom collapsed, capital vanished and equipment makers learned that technical merit could not outrun a broken market. By March 2004, Photuris had exhausted its funding and sent its staff home.
Vengsarkar tried to make the ending less wasteful. He argued that the products only made sense if a buyer took some of the people who had developed them. The assets were eventually sold. He later summarized the period plainly: the company went bankrupt, but the team managed to preserve jobs. Failure entered his story as an operating fact, not an inspirational costume.
The next company carried the lesson. Nistica, founded in 2005 by veterans of Photuris and Bell Labs, focused on wavelength-selective switching modules. Instead of becoming its own captive customer by building a complete system, it sold subsystems to multiple equipment makers. Instead of keeping every manufacturing step inside, it planned to outsource early while retaining final assembly. The product strategy changed because the postmortem had been specific.
Nistica entered a strategic partnership with NTT Electronics in 2010. Fujikura invested and later acquired the company, while its founders and key employees continued developing optical products. In 2018, Molex acquired the Nistica business and folded it into its Optical Solutions Group. Vengsarkar said the larger platform would help the team expand capacity and bring a broader portfolio to market. The arc ran from an intentionally lean startup model to industrial scale.
Rewiring for Google scale
At Google, the unit of thought changed. Vengsarkar joined the group that develops, builds and operates the company’s global fiber network. Individual optical streams ran at 400 gigabits per second; the overall system carried traffic at petabit scale. Speed mattered, but reliability, cost and power mattered beside it. A change in one part could force adjustments throughout an ecosystem of equipment and software.
He described the transition as rewiring his brain for a global mindset. Learning began with listening and with technical documents, followed by the work of explaining complexity to other people. His planning horizon stretched five or ten years. Vendors and university researchers became part of a continuing conversation about which ideas might change the network rather than merely tune it.
Here the crossword parallel becomes useful. A constructor cannot optimize one answer in isolation. A gorgeous entry that wrecks the surrounding grid is not gorgeous for long. Vengsarkar’s account of networking sounds similar: higher line speeds arrive, then every adjacent component and management system must respond. The achievement is coherence.
Back into the crucible
Vengsarkar joined nEye.ai as CEO in 2025. The company had grown out of more than a decade of research in Professor Ming Wu’s laboratory at UC Berkeley. Its core proposition was an optical circuit switch on a chip, combining silicon photonics, microelectromechanical systems and CMOS. The switch could alter how processors and memory connect without repeatedly converting the traffic through power-hungry electrical switching layers.
The move also returned him to a familiar professional network. nEye’s vice president of engineering, Jefferson Wagener, had worked with Vengsarkar at Bell Labs and co-founded Nistica. Their names share patents from the 1990s, including work on optical gratings. At nEye, they joined a Berkeley-born founding group that includes chief scientist Ming Wu, chief technology officer Tae Joon Seok and co-founder Kyungmok Kwon. The combination is characteristic of long technical careers: new companies often assemble from trust built on earlier, exacting work.
AI made that proposition timely. Training and inference clusters were expanding faster than the copper and electronics linking them. The industry’s familiar story centered on the processors, yet expensive compute can sit idle when data cannot move to the right place quickly enough. nEye’s wager sits in that gap: change the topology of a cluster on demand and use light to reduce the burden of moving information.
The investor list connects Vengsarkar’s past to the company’s intended market. CapitalG, Alphabet’s independent growth fund, led a $58 million Series B announced in April 2025. M12, Microsoft’s venture fund, also invested, alongside semiconductor and infrastructure names. In July, nEye joined the Open Compute Project’s Open OCS effort with Google, Nvidia and Microsoft.
Those connections do not remove the commercial risk. They do make the feedback loop unusually direct. The organizations backing or collaborating with nEye operate the sort of large computing systems the product is meant to serve. Their interest gives the company access to informed questions about topology, density, power and operations. Vengsarkar’s role is to turn that attention into a product program without confusing industry need for guaranteed adoption.
nEye’s capital base, by announced cumulative total
In April 2026, Sutter Hill Ventures led an $80 million Series C, bringing nEye’s announced funding to $152 million. Vengsarkar’s response concentrated on what comes after validation: foundry-based manufacturing and demanding customer performance standards. His shorter version was sharper: “It’s time to move from the lab to the fab!”
That sentence contains the less cinematic half of deep tech. A prototype must become a repeatable product. The company has to manage yield, packaging, testing, supply, reliability and cost while the target market keeps changing. Vengsarkar has already lived versions of that transition, including one company that could not finish it and another that reached it through strategic buyers.
The other grid
Crosswords remain his counterweight. He has said that he loves words and constantly produces anagrams. Once, on hearing a new colleague’s name, he immediately rearranged it into “adventure.” A seed entry for his 2021 Sunday puzzle “No Ruse” arrived while he was sitting at the dentist. He also plays raga and table tennis; crossword editor Will Shortz has mentioned their annual rivalry at the table.
His nonprofit work widens the picture. As secretary of Maker Bhavan Foundation, he supports hands-on STEM education. He has also participated in high-school science fairs and German Shepherd rescue work. These are quieter forms of systems building: make a place where curiosity can become practice, or help a network of volunteers protect animals.
nEye now faces a grid with more crossings than any puzzle. Foundries, hyperscalers, investors, standards groups and engineers all need answers that agree. Vengsarkar brings a memory of the telecom crash, the discipline of component manufacturing and the scale of Google’s network. He also brings the constructor’s habit of looking at the whole pattern before committing to one square.
The work ahead will be measured in production performance, not wordplay. Yet the sensibility travels. Complex systems rarely yield to one dramatic move. They yield when enough careful connections hold at once.