Sep 26, 2026 · 32 min · 10 segments
**EPISODE DESCRIPTION** I sat down with Rafsun, founder of Sahaj Mobile, at Money2020 to unpack one of the most exciting fintech stories I've come across in a long time. Rafsun went from managing…
Sam KamaniHost
And even the VC, the family office guy I was interviewing, asked him the same question.
Yeah.
Basically, I think from the moment I wake up, To
24-7.
Even now, you know, I think they're calculating our loss reserve and, like, figuring out how to, you know.
So from a personal perspective, I am just all in it, you know? Yeah.
Not so much on a, like, it's...
It really augments my ability to do tons of work.
Because of, yeah, I don't have like a assistant that would do a lot, like meeting stuff like that.
The review process now, the iteration cycle has just gotten better and better.
That's what I'll say from my end.
But as far as our company goes, it's completely embedded.
came from running a machine learning model that says, hey, let's look at correlation between your transaction history and how much you defaulted, okay? So it's very, you know, At a place like South Asia and Bangladesh, there's no credit spares, right? So we have to rely on alternative data to kind of go through, to understand the repayment behavior.
Locking helps, but we don't want At the outset, we want to filter out the best customers.
In order to do that, we use massive systems to run models, and these are huge data sets.
These take days of running in order to come to a solution.
From that, we're doing tremendous amount of work.
Third thing is on the customer service front, chatting.

And even the VC, the family office guy I was interviewing, asked him the same question.
Yeah.
Basically, I think from the moment I wake up, To
24-7.
Even now, you know, I think they're calculating our loss reserve and, like, figuring out how to, you know.
So from a personal perspective, I am just all in it, you know? Yeah.
Not so much on a, like, it's...
It really augments my ability to do tons of work.
Because of, yeah, I don't have like a assistant that would do a lot, like meeting stuff like that.
The review process now, the iteration cycle has just gotten better and better.
That's what I'll say from my end.
But as far as our company goes, it's completely embedded.
came from running a machine learning model that says, hey, let's look at correlation between your transaction history and how much you defaulted, okay? So it's very, you know, At a place like South Asia and Bangladesh, there's no credit spares, right? So we have to rely on alternative data to kind of go through, to understand the repayment behavior.
Locking helps, but we don't want At the outset, we want to filter out the best customers.
In order to do that, we use massive systems to run models, and these are huge data sets.
These take days of running in order to come to a solution.
From that, we're doing tremendous amount of work.
Third thing is on the customer service front, chatting.
The rest of this transcript — segmented and speaker-labeled, so you land on the exact moment something was said
Search every transcript — by keyword, by phrase, or by meaning, across every show Radar indexes
Trends — what is surging across podcasts, measured against its own baseline
Alerts — when a name you follow appears in a newly indexed episode
No account is needed to search Radar.