The team behind Panthera spent years inside credit operations at Asian digital lenders before deciding the tooling problem wasn't going to fix itself. We founded Panthera in Tokyo in 2024 to build the scoring infrastructure we needed and couldn't find.
Spent six years building credit scoring infrastructure at consumer lenders in Vietnam and Singapore before founding Panthera in Tokyo in 2024. He saw thin-file rejection rates firsthand at every institution he worked at, and what those rejections meant for the borrowers on the other side. Kien leads Panthera's market strategy and lender partnerships.
Built alternative-data feature engineering pipelines at a Jakarta-based neobank before co-founding Panthera. Her work on gig-income volatility patterns is the foundation of the Indonesia market model. Priya leads all model architecture, feature extraction, and inference infrastructure at Panthera.
Ran business development at a Tokyo credit union for a decade before moving into fintech advisory. Takeshi brings the lender-side perspective that keeps Panthera's output format and reason codes aligned with what underwriters actually need. He leads data source partnerships and regulatory engagement with FSA and FISC for the Japan market.
Bureau-based credit scoring was designed for formal employment economies with centralized financial infrastructure. Most of Asia's economically active population doesn't fit that model. They work in informal or gig arrangements, hold income in mobile wallets, and clear obligations through digital platforms that traditional bureaus have never touched.
The problem isn't that these borrowers are high risk. The problem is that existing models have no signal on them at all. A lender looking at a thin-file applicant without Panthera has two choices: reject without review, or lend blind. Neither is good lending.
We don't try to replace bureau data. We provide the signal layer for the population where bureaus have nothing. That's a narrower scope than competitors who try to do everything, and it lets us build models that are genuinely accurate for thin-file borrowers rather than a general-purpose approximation.
Founded 2024, bootstrapped to first production deployments
Headquarters Ark Mori Building, Akasaka, Tokyo 107-0052
Active markets Vietnam, Indonesia, the Philippines, and Japan
Model update cadence Quarterly retraining against live repayment outcomes
Average approval lift +38% on thin-file applicants at same default threshold across pilot deployments
API latency P99 under 800ms including feature extraction
Regulatory documentation Provided for SBV, OJK, BSP, and FSA requirements at every deployment
If you're an OJK, SBV, BSP, or FSA-licensed lender looking to serve borrowers your current model can't reach, we'd like to hear about your market, your data environment, and what pilot results would look like for you.
