Hackathon Participants Solve Real-Time Reconciliation Problem In 36 Hours; Procurement Indicates Build Will Take 14 Months

The winning team at this year’s annual Australian fintech hackathon, held over a long weekend in Sydney, produced a working prototype that resolves a long-standing real-time reconciliation problem affecting several mid-tier payments firms. The prototype, developed in 36 hours by four engineers and a product designer, was demonstrated live on the closing day. Industry observers described the demonstration as impressive. The procurement function of the host institution has indicated the production build will take 14 months and will require a different team.

The hackathon, now in its sixth year, attracted approximately 380 participants and 47 sponsoring organisations. The theme was “Reimagining Operational Processes For The Real-Time Era”. Participants formed teams of between three and six people and were given a fixed problem statement: identify and prototype a solution to one of seven nominated industry pain points. Of the seven, the reconciliation problem was selected by the winning team because, in the team’s words, “it looked solvable”.

The team comprised an engineer from a tier-two bank, two engineers from a regtech, a product manager from a payments processor, and a designer who had attended in a personal capacity. The team had not previously worked together. The team had not previously met. The team divided the problem into three workstreams over the first afternoon. The team produced a working prototype by the second evening.

The Prototype

The prototype, demonstrated to a panel of judges drawn from across the industry, addresses a reconciliation issue that has been documented in industry working group papers since 2019. The issue affects daily reconciliation across multiple settlement windows, and is well known among operations teams. Several attempts to resolve it through formal vendor processes have been initiated. None has been completed.

The prototype consists of a small service that ingests two existing data feeds, normalises them in flight, and outputs a reconciled record set. The judges said the architecture was “elegant” and the implementation was “clean”. The team provided full source code, documentation, and a deployment guide. The deployment guide is one page. It contains the four commands required to run the prototype.

The team estimated that, with appropriate testing, the prototype could be deployed in production within six to eight weeks. A senior compliance professional present at the demonstration said the timeline was “optimistic but not unreasonable”, and that the prototype appeared to be of a standard consistent with the institution’s production code. The compliance professional did not have any further comments at that time.

“This is exactly the kind of innovation our industry needs,” a senior executive at the host institution said in remarks at the closing ceremony. “We will be looking very closely at how we can take learnings from this work into our broader transformation programme.” The transformation programme commenced in 2022. It is currently in its planning phase.

The Production Pathway

Following the closing ceremony, the host institution’s procurement and vendor management functions confirmed that the prototype could not be deployed in its current form. The institution’s procurement framework requires that any production solution be procured through a competitive request for proposals, with at least three vendors invited to participate. The hackathon team is not a vendor. The hackathon team is four people who happened to be in the same room for a weekend.

A spokesperson for the procurement function said the institution would now “assess the appropriate procurement pathway” for any future production solution addressing the reconciliation problem. The assessment will be conducted over four to six weeks. It will be reviewed by the procurement steering committee. The procurement steering committee meets monthly. The next meeting is in three weeks. The agenda for the next meeting is full.

The institution further confirmed that any production build would require a separate vendor selection process, an architectural review, a security review, a privacy impact assessment, an enterprise change advisory board approval, and a go-no-go decision at executive committee. The total estimated time to delivery, assuming all approvals proceed without exception, is 14 months. The total estimated cost is between $2.4 million and $3.1 million. The hackathon prize is $20,000.

The institution noted that the hackathon team’s source code would be “retained for reference”. The team will not be invited to participate in the production build. Two members of the team have indicated they would not have been available regardless. One of them works for a regtech that has subsequently filed an expression of interest in the request for proposals.

Industry Reception

Industry observers described the outcome as consistent with broader patterns. A consultant familiar with the institution’s technology procurement said the gap between hackathon outputs and production deployment was “well understood”. The consultant noted that, of the past five hackathon-winning prototypes, four had not been deployed in production. The fifth had been deployed in production after a two-year procurement process, with a different vendor, using a different architecture. The original team had not been involved.

A senior payments executive at a peer institution said the pattern was “a recurring source of frustration” for innovation teams. “We celebrate hackathons,” the executive said. “We applaud the energy. We then return to the regular processes. The regular processes are designed for stability, not speed. Both are valuable. Only one is celebrated.”

The host institution confirmed that next year’s hackathon would proceed as planned. The theme for next year is expected to be “Reimagining Operational Processes For The Real-Time Era”. The reconciliation problem is expected to be one of the nominated problem statements. The institution is not planning to deploy the current solution in advance.