Built after watching a team burn a month closing the books on spreadsheets
HyperCal started in Seoul in 2025 when a small team of actuarial engineers decided the quarterly close cycle for insurance carriers did not have to be what it was.
Why we built this
Kim Sung-jun spent seven years building actuarial software for Korean insurance groups. Every quarter, without exception, the reserving team would start over. Development triangles rebuilt in Excel. Methods reselected, sometimes inconsistently with the prior quarter. Tail factors adjusted in cells nobody else could find. The audit trail reconstructed from memory after the fact, if the regulator asked.
The problem was not that the actuaries were doing it wrong. They were doing it correctly, given the tools they had. The problem was the tools. Every insight, every calibration, every defensible assumption existed only in the spreadsheet and in the actuary who built it. When that person left, or when a regulator asked a question two years later, the answer was not in any system.
HyperCal runs the same actuarial methods your team trusts. The difference is that every assumption, every run, and every approval is captured automatically. The audit trail is not reconstructed after the fact. It is generated as part of the model run itself.
The people behind HyperCal
Three practitioners who spent years inside the problems we are solving.
Built actuarial software for Korean insurance groups for seven years before co-founding HyperCal. Deep background in reserve cycle engineering and regulatory reporting pipelines. His work spanned property, casualty, and health lines at carriers across the Korean market.
Spent years building high-throughput data pipeline infrastructure for financial services applications. His background spans distributed systems engineering and financial-grade data integrity. Designed the immutable audit log architecture that underpins every HyperCal model run.
Actuarial modeling career at Korean non-life insurers spanning reserving, pricing, and regulatory capital work. Brings practitioner knowledge of where manual processes introduce risk. Responsible for validating every HyperCal method implementation against established actuarial standards.
Principles we build by
Actuarial methods, not replacements
We implement Chain Ladder, Bornhuetter-Ferguson, Cape Cod, and frequency-severity methods as actuaries understand them. We are not trying to replace actuarial judgment with a black box. We are automating the rebuild so judgment can be applied to decisions, not to formula maintenance.
Regulator-first documentation
Every output we produce is designed to be placed in front of a regulator without further assembly. The audit trail is not a feature you turn on for examinations. It is the default output of every model run, structured so a financial examiner can follow the chain from assumption to final reserve figure.
Honest about what we are
HyperCal is a small team in early access. We do not claim to have hundreds of carrier deployments or a decade of track record. We work closely with each carrier in our program, and we do not grow beyond what we can support well. When we tell you something works, it is because we have seen it work on real reserve cycles.
Data stays in Seoul
All customer data is stored in South Korea. We do not replicate to foreign jurisdictions by default. For carriers subject to Korean insurance regulation, keeping data in-country is not optional. We built our infrastructure with that requirement as the starting constraint, not an afterthought.