Thirty-one years of building, running and replacing systems that move money

Mandate

Complete the move to the Kaname shared core, then bring the bank's recovery capability, cloud use and payment services into line with the new operating model.

Outcomes

• Completed the migration of 3.4 million accounts to Kaname over Golden Week 2025, after a four-month postponement costing ¥2.3bn.
• Reduced measured restore time for critical non-core systems from 61 hours in 2023 to 9 hours 40 minutes in 2025.
• Moved the analytics platform to public cloud in October 2025 after a rehearsed exit that took 58 hours against a 72-hour limit.
• Closed the bank's own payment app in March 2026, saving about ¥380m a year.
• Cut privileged administrative accounts from 388 to 74.

Decisions in context

My appointment as chief technology officer added information security and digital channel engineering to the Systems Division I already led, and gave me a seat on the executive committee. The first two years were dominated by the migration: the decision to postpone in November 2024, the two clean rehearsals that followed and the cut-over over Golden Week 2025. Around it I commissioned the 2023 restore test that exposed how weak our recovery was, and moved ¥1.3 billion into isolated backups at the cost of delaying a project I had championed. Since the migration I have turned to the questions a shared core leaves each member with: which services we must be able to run ourselves, how far to use public cloud and which payment services to keep. I report to the president, present system risk to the board and report to the audit and supervisory committee on audit findings and incidents. I am not a director of either company.

Reason for the move

Current appointment; completing the Kaname resilience programme and the cloud move of internet banking.

Mandate

Contract with the Kaname consortium and Mikage Information Systems, run the migration programme and keep the old mainframe stable until cut-over.

Outcomes

• Signed the Kaname participation agreement in March 2020 and reduced 211 bank-specific customisations to the 23 the consortium accepted.
• Introduced migration-specialist contracts in 2021; 11 of 12 re-employed engineers stayed until after cut-over.
• Started the business-rule library with the owning divisions; it reached 2,140 recorded rules, 318 previously undocumented.
• Ran the mainframe from 2020 to 2023 without a customer-facing outage longer than thirty minutes.

Decisions in context

This was the role in which the migration became a programme rather than a recommendation. I negotiated the participation agreement with the Kaname member banks and Mikage, and led the hardest early work, which was not technical: persuading our business divisions to give up 188 of the 211 customisations they regarded as essential. Some were genuinely valuable and we argued for them in the consortium's committees; most existed because nobody had asked whether they were still used. At the same time we had to keep a thirty-year-old mainframe stable while the people who knew it best approached retirement, which led to the migration-specialist contracts and the business-rule library. I reported directly to the president, who chaired the programme's steering committee, and presented progress and risk to the board each quarter. I also learnt how to work inside a consortium, where every change has eight other banks' views attached to it.

Reason for the move

Appointed managing executive officer and chief technology officer, with information security and digital channels added to the remit.

Mandate

Recommend what should replace the bank's own mainframe before hardware support ended in 2024, and modernise technology planning.

Outcomes

• Compared rebuilding the core with three shared-system options over eighteen months and recommended Kaname; board approval in December 2019.
• Disclosed prior employment with the incumbent vendor and stood aside from scoring its proposal.
• Built a ten-year cost model showing core running costs about 30% lower in a shared system.
• Delivered the bank's smartphone payment app for its 2019 launch.

Decisions in context

Shiomine hired me from outside regional banking to answer one question: what should replace the mainframe? The study took eighteen months. Rebuilding our own core on modern technology would have cost about 40% more over ten years and left us carrying the full burden of change. Of the three shared systems we examined, Kaname had the strongest operator and the most rigorous change governance, and the least room for our customisations. I recommended it knowing that our business divisions would feel the loss of flexibility for years. Because Kogetsu, my former employer, ran our existing system and bid for the replacement, I disclosed that history at the outset and stood aside from scoring its proposal. In the same period my team delivered the smartphone payment app the president wanted ahead of the bank's centenary in 2021, a project I later recommended closing. The role taught me to explain technology choices to directors as choices about what the bank would give up.

Reason for the move

Promoted to executive officer to lead the migration programme and the Systems Division.

Mandate

Replace an ageing authorisation system and reduce fraud losses while card volumes grew.

Outcomes

• Rebuilt authorisation to run in two data centres at once, with peak capacity of 2,400 requests a second.
• Introduced real-time fraud scoring; annual fraud losses fell from ¥1.24bn to ¥0.71bn while transaction value grew about 40%.
• Raised the share of IC-capable terminals at acquired merchants from 31% to 88% between 2013 and 2016.
• Ran authorisation without a full outage from 2013 to 2016.

Decisions in context

I moved to Seiran to work at larger scale and in a part of finance where failure is measured in seconds. Card authorisation must answer millions of requests a day in well under a second, and a pause at a supermarket checkout becomes a customer complaint within minutes. The system I inherited ran in one data centre with a standby that had never carried live traffic. We rebuilt it to run in two centres simultaneously, so that losing one reduced capacity rather than stopping service. Fraud was the second mandate. Real-time scoring cut losses substantially, though each new control produced a new pattern of attack within weeks, which taught me to treat fraud defence as continuous work rather than a project. On the acquiring side I saw payments from the merchant's counter, as fees, settlement timing and terminals. From 2014 I was an executive officer and presented fraud and resilience reports to Seiran's board each quarter.

Reason for the move

Recruited by Shiomine Bank to lead the planning for replacement of its core banking system.

Mandate

Stabilise and extend a newly opened bank's open-systems core, then run its operations and security.

Outcomes

• Scaled the core from launch to 1.3 million accounts and ¥1.1 trillion of deposits.
• Introduced transfers between Hanasato accounts at any hour, including weekends, in 2004.
• Led the response to a seven-hour transfer outage in March 2008 and rebuilt failover around rehearsals under month-end load.
• Moved all customers to one-time passwords for transfers by 2009.

Decisions in context

Hanasato opened in August 2001 with a small team, an open-systems core and the ambition to let customers bank at any hour. I joined two months later to lead core systems development, and it was my first experience on the client side of a contract. Much of the early work was catching up with growth: account numbers rose faster than the original design assumed, and we rebuilt the batch and the database layer twice in five years. From 2006 I ran IT operations and security. The outage in March 2008 was the defining event. Failover that had passed eleven tests failed under real month-end load, and 41,000 transfers were delayed for up to seven hours. I reported it to the Financial Services Agency, rebuilt our testing around realistic load and learnt to distrust any test performed under convenient conditions. We also moved all customers to one-time passwords as fraud against online banking grew.

Reason for the move

Moved to Seiran Card Services to lead payment systems at larger scale.

Mandate

Develop and maintain deposit and loan applications for regional bank clients, then lead year-2000 and channel projects.

Outcomes

• Wrote and tested interest-calculation programs for Shiomine Bank's deposit system on site in Niigata in 1997.
• Reviewed date logic in about 1.2 million lines of code for two banks' year-2000 remediation.
• Led a 14-engineer team that moved a regional bank's ATM network to a new interface in 2000 without customer-facing downtime.

Decisions in context

I joined Kogetsu as a graduate engineer in April 1995 and learnt COBOL on deposit and loan applications it maintained for regional banks. The work was detailed and often repetitive, and it was the best possible education in how a bank's rules become a system. In 1997 I spent eleven months on site in Niigata writing and testing interest-calculation programs for Shiomine Bank, unaware that I would retire them as an executive in 2025. The year-2000 work in 1998 and 1999 was my first experience of a deadline that could not move, and of how much undocumented logic sits inside old programs. As a project leader from 1999 I ran a fourteen-person team that moved a regional bank's ATM network to a new interface over a holiday weekend. I left in 2001 because I wanted to see banking technology from the client side, where the consequences of a decision stay with you.

Reason for the move

Joined a newly opened internet-only bank to work on the client side and on open systems.