Cloudera touts hybrid data platform to power enterprise AI boom

Cloudera is pitching itself as the essential “shovel to the miners” for the enterprise artificial intelligence (AI) gold rush, as it sees a future where customers no longer have to choose between the control of on-premises infrastructure and the convenience of the public cloud.

Speaking at the company’s Evolve Asia-Pacific (APAC) event in Singapore this week, Cloudera CEO Charles Sansbury noted that more enterprises are pushing back against all-in cloud migrations to retain control over their most sensitive data, particularly for AI workloads.

“Two years ago, people would tell you how quickly they’d get all their workloads to the cloud. Now they’re arguing about how many of their workloads will stay on-premises,” Sansbury said, citing research suggesting that 40% of workloads are expected to remain on-premises – a figure he expects will grow.

He claimed that Cloudera is the only company capable of delivering a hybrid data platform for analytics and AI across different computing environments, which he argued is becoming critical for modern data architectures. That was also why the company recently acquired Taikun, a platform for managing Kubernetes and cloud infrastructure across hybrid and multi-cloud environments.

Cloudera’s chief product officer, Leo Brunnick, and chief technology officer, Sergio Gago, framed the Taikun acquisition as the “missing piece” to deliver on the company’s vision. They said it gives customers a single reference architecture with one codebase and user experience, eliminating the distinction between public and private cloud offerings. This provides a seamless experience for data engineering, AI inferencing, and data visualisation, regardless of where the data or compute resides.

A key customer voice validating Cloudera’s data platform came from Donald MacDonald, managing director and head of the group data office at OCBC Bank. While the bank’s data team started with customer analytics and marketing for its AI use cases, MacDonald wanted to tap bigger opportunities for transformation in “material areas” such as risk and compliance.

That was when the team decided to take on a high-stakes project in financial crime compliance. “We were using traditional rules-based systems for anti-money laundering,” MacDonald said. “We were producing 12,000 alerts every month. Every alert took 40 minutes to investigate and 98% of those alerts were false positives. We had hundreds of people across the business doing low-value work.”

By applying AI to score alerts with an additional 500 features, OCBC was able to automatically handle low-risk alerts and redeploy hundreds of staff to higher-value tasks. “The business could see that the data team are not just working on marketing use cases,” MacDonald said. “They’re willing to take on material things where there’s a lot of risk at stake.”

MacDonald attributed the bank’s ability to scale its AI efforts to three key pillars: people, platform and data. Central to the data and platform pillars is its use of Cloudera to build a data lake and an enterprise data science platform in a private cloud environment.

“Good AI doesn’t happen without good data,” he said. “For the last 10 years, that's been in Cloudera, where all of our data is centralised. We have more than 350 systems now sitting in Cloudera, updated daily, 20 of them in real-time.”

OCBC’s unified data foundation, combined with an in-house machine learning operations (MLOps) framework and an AI centre of excellence, allows it to drive AI projects and reuse data products across the organisation. The bank’s heavy use of automation has also saved an estimated 25-30% of its data scientists’ time, enabling it to do more with less.

Frank O’Dowd, Cloudera’s chief revenue officer, noted that the APAC region has been a hotbed of innovation from customers like OCBC. “There is no market that exhibits this better than APAC,” he said. “I share stories from things that you’re doing all around the world.”