From Cary to Pune: An analytics titan is forging its AI innovation future

Time of India

07-05-2025

representative image (ANI)
Bryan Harris said:
Expanding responsibilities
Everything about analytics software firm SAS is unassuming -like an iceberg that hides its sheer size beneath the waters. It started out as a research project at North Carolina State University in the 1960s to create a statistical analysis system, was incorporated in 1976 and has had the same CEO for the last 49 years, James Goodnight. It has been routinely praised for decades for using higher proportions of its revenue on research and development as compared to its peers. It is also one of the largest privately held software companies in the world. Suffice it to say, SAS Institute , as it is officially called, has seen all the tech waves, bubbles, and slumps imaginable and is still relevant today precisely because of its focus on R&D.Now SAS finds itself in the midst of another tech revolution – artificial intelligence – and the company's Pune R&D centre, set up 25 years ago, is playing a central role in helping the company stay ahead of the curve. That's what the firm's CTO Bryan Harris told us during his recent visit to Pune.'Pune is our largest remote R&D centre in the world, and it's an absolutely strategic asset for us,' Harris declared. 'Every part of our portfolio— our Viya platform (SAS' artificial intelligence, analytics, and data management platform) our industry solutions and our stand-alone AI models—has the fingerprints of Indian engineers on it.'The company employs just over 12,000 people worldwide, and more than a thousand of them now sit in India; about 800 are hard-core developers. 'The site has been growing 30 percent a year, and I expect it to keep that pace for the foreseeable future,' Harris said. 'Pune already looks like a miniaturised version of our headquarters in Cary, North Carolina.'The SAS of today is pitching a three-layered AI stack: build on Viya if you have data scientists, buy off-the-shelf solutions if you don't, or plug in individual models to finetune existing workflows. 'Whether a bank wants a full customer-intelligence suite or just a single supply-chain model, we give them the choice,' he said. 'And Pune is building code for every one of those choices.'Take Customer Intelligence 360 , the flagship marketing product that watches a website's clickstream data (a record of all the links and actions a user takes while navigating a website) and decides, in real time, whether to tempt a visitor with a creditcard teaser or a home-loan refinance. 'We're very highly regarded for CI 360, and the Pune team has been instrumental in delivering it as a service and accelerating innovation,' Harris noted.The same engineers keep the tills ringing in banking. 'This centre is the hub for our fraud and risk solutions—capital adequacy, regulatory compliance, instant card-swipe fraud detection,' he said. One neural-network model used by global banks tips the scales at 18 terabytes uncompressed yet must rule on a transaction in under 30 milliseconds. 'That model has evolved here for decades,' Harris added with pride.India's importance is heightened by SAS's unusually balanced revenue mix. 'Half our business is outside the Americas—that's very rare for a US software firm,' Harris observed. 'If we need to serve 95 percent of the world's top banks, we must meet them where they operate, and India is an essential multi-functional hub for that.'The growth spurt is not merely about headcount but about autonomy. Harris spent much of his Pune visit mapping the next wave of responsibilities the site will own. 'I want the engineers here making strategic design decisions, not just writing code to a spec,' he said. 'They already run global network operations and technical support; now we're pushing deeper into platform architecture and model governance.'Generative AI may dominate headlines, but Harris insists that traditional, deterministic models still drive most profits. 'Large language models are marvellous, but they don't issue a mortgage or spot card fraud in 30 milliseconds,' he said. 'The magic lies in orchestrating generative and deterministic models together—and our intelligent decisioning application, again built with heavy lifting from Pune, does exactly that.'