Indeed puts AI to work to help job seekers find new roles

Indeed’s cloud-native and data-driven digital transformation has set it up ideally for the rapidly advancing AI era, says Anthony Moisant, the company’s CIO and CSO.

“We’re not just adopting AI. We’re building an agentic company where human intelligence and empathy can come together with machine intelligence,” Moisant says.

“What this tremendous technology transformation has given us is better connection with job seekers, employers, and our own employees because we’ve stripped a lot of the friction out of the gear,” he adds. “That makes for richer conversations with people.”

Moisant, who also serves as executive officer of parent company Recruit Holdings and previously led IT at sister company Glassdoor, knows a bit about empathy.  

A graduate of Boston University’s CIO Executive Program and the US Navy Nuclear Submarine School, Moisant developed his love for IT and coding while serving aboard a nuclear submarine. Moisant — who found himself homeless before joining the military — emphasizes that his heart is as committed to Indeed’s charter as his IT IQ.

He was in Italy when Glassdoor reached out to him about an opening leading IT at the company. “I jumped at the chance to join Glassdoor when it was pretty small,” Moisant says. “I fell in love with this idea of helping people get jobs. Being jobless, without economic security, really resonated with me.”

Setting the foundation for innovation

When he arrived at the company more than five years ago, much of the infrastructure was built on a legacy, on-premises server platform architecture that the IT team believed was more cost effective and gave them greater control of the company’s destiny.

At the time, it was rare for a company of its size not to be running workloads in the cloud, but its microservices platform, Moisant knew, would be an asset when the time came to migrate. Knowing the cloud would be a game changer for Indeed, Moisant began moving workloads to AWS to achieve immediate agility gains while preparing the IT staff for a new path forward.

Planned roughly four and a half years ago, Indeed’s migration to AWS took about a year and a half to complete. Initially a lift-and-shift approach, Moisant and the IT team focused the next year on transforming Indeed’s workloads to be cloud native. By embracing openness, container orchestration, and API-first event driven design, Moisant knew a loosely coupled cloud-native architecture would not only enable Indeed to maximize the value of its applications on the cloud but also position the company well for the AI era.

“The first pivot was moving to become an agile organization, getting into the hyperscaler model, pivoting our services toward that, and unifying our data strategies to get ready for the next wave of transformation,” Moisant says.

Overhauling the company’s data architecture was a top priority. First, Indeed moved data lakes from its on-premises environment to AWS. The company’s data was stored in Snowflake, Tableau, and various Elastic MapReduce (EMS)-based technologies, all of which Moisant and team moved to a unified data lake and lake house on AWS. Doing so established a shared data architecture that could scale and be more flexible, not just for analytics use cases but also operational use cases, the CIO says.

With an eye toward the AI future, Moisant wanted an open data architecture that would enable observability, governance, data classification, and quality control. He and his team opted to build Indeed’s data lake house on Apache Iceberg.

That decision delivered far more value for Indeed than expected, as industry dynamics unfolded. “We wanted to be engine-agnostic and unified,” Moisant says. “Essentially, the query engine, like whether we use Snowflake or not, became less relevant. It was more relevant to have a unified infrastructure.”

He adds, “So when Salesforce and Amazon and Google and most of the big players out there started making it easier to use Iceberg transparently, it happened to unlock a lot more incremental value. I’d like to say we picked that strategy for interoperability with all these cloud providers, but we didn’t. We chose Iceberg because it was more durable, performant, and flexible.”

A job for AI

Indeed engineers have been developing machine learning (ML) algorithms and models for many years to create the best matches between job seekers and employers. But Moisant and his IT team thought generative AI was much further out when they started Indeed’s migration to the cloud nearly half a decade back. Still they started “dancing” around generative AI pilots roughly three years ago that were “frontier” and “promising,” he says.

The team’s focus on data quality made production models possible two years ago, and shortly thereafter, Moisant made the decision to deploy AI everywhere.

The CIO’s plan involves training OpenAI-based models for now, but the company intends to remain model-agnostic and will choose whichever model — large or small — to achieve what Moisant describes as a company-wide effort to identify new ways of matching prospective employees with vendors and “reimagining” how people get jobs, he says.

Much of the generative AI and agentic AI development at Indeed has been and will be built in-house by using AWS “Bedrock-type” platforms and AI-based developer tools, Moisant says.

 Another core strategy for Indeed is deploying Salesforce agents, primarily for go-to-market activities.

“We don’t have to build our own agentic solutions for go-to-market sales and customer service needs,” Moisant says. “Salesforce is a platform play and what we have acquired is speed and quality. We get to leverage their innovation for those cases and that allows us to focus on differentiation in our products.”

Indeed currently has four Salesforce agents in production and six internal products that the CIO claims are agentic. More than 20 other agents are in development as well.

Agentic AI will likely alter and fully automate many of Indeed’s key business processes, but Moisant is clear that human input and human interaction will always have a role to play. “At the end of the day, we are mission-driven,” the CIO and CSO says. “Everything ties back to helping people get jobs faster — and that is human.”