Thought Leaders in Artificial Intelligence: Ashutosh Garg, CEO of Eightfold.ai (Part 2)

Sramana Mitra: Let me ask you a few questions to clarify how your algorithm is working. What is the database of candidates that you’re working with?

If it’s a large enterprise, are they starting with a database of candidates that exist within the enterprise, and are you supplementing that with data from some other sources? What is the base database?

Ashutosh Garg: The base database comes from the enterprise. This is the database of each and every person they have ever engaged in any capacity. 

Sramana Mitra: The same goes for recruiters. It’s their own database. Your algorithm starts life on the customer’s database.

Ashutosh Garg: That’s right.

Sramana Mitra: On what parameters are you tagging each candidate? What are the different types of parameters that you’re checking on?

Ashutosh Garg: We are evaluating the fit of a candidate for a job. We take hundreds of features into account. We try to understand who has performed well in this job while removing any bias that is in the organization. Then, we match the people based on that.

There is a candidate who is a software engineer for the front end, has worked for companies on the consumer side, it’s quite likely that this person can become a product manager. Our system will take this and match it to a product manager role.

Sramana Mitra: Talk about some of the software parameters that are driving your algorithm. If you’re trying to categorize on the basis of culture fit, that is a whole different level of deciphering signals. 

Ashutosh Garg: That’s a very good question. Whether they have been in a customer setting role or not, or whether they have been in the role of managing the customers, we try to understand those things.

We also try to understand what kinds of people enterprises hired in the past – whether these are people who are risk takers, people who have shown high learning ability, people who are confident, and perform well in a structured environment.

We try to infer those things based on someone’s experience and use that to match them to a role. Let’s say, as a startup, you’re trying to hire someone. If you see that there is someone who has been in a large company for the last 10 years, has been in one team, and focused on this one technology, it’s unlikely that this person will perform well in a startup.

If this person has moved across different number of roles in the same organization, has been active in terms of publishing like blogs, they are likely to perform well in a startup.

Sramana Mitra: Can you give more examples? This is a slightly more complex and involved issue. I’d like to have a few more examples to illustrate how you draw the signals. 

Ashutosh Garg: I did my profile. Based on that, what came across was that I’m a very risk-averse person.

Sramana Mitra: But you’re a company founder.

Ashutosh Garg: That is the issue. What we are saying is that this person was in India, finished undergrad, travelled to the US, left Google when the market was tumbling down to start something. He will have risk-taking ability.

Quite a few times, people who have done PhD and have done a lot of publications have very strong training on public speaking side because they’re constantly going to conferences and presenting their research. They will never put in their profile that they know public speaking.

If you are part of an enterprise or a strategic sales team of a large organization selling large ticket items, you’re probably very good in negotiation. If you are a product manager in a high-growth company, you’re probably a very analytical person. The system is trying to infer these kinds of insights based on the profile.

This segment is part 2 in the series : Thought Leaders in Artificial Intelligence: Ashutosh Garg, CEO of Eightfold.ai
1 2 3 4 5 6