An award-recognized banking technology innovator who modernizes ACH, Wire, and SWIFT systems at one of America's largest financial institutions spent 72 hours evaluating hackathon projects built around intentional failure -- and found that the principles governing trillion-dollar payment networks apply directly to software designed to break.
The Automated Clearing House network processed $76.7 trillion in 2023 across 31.5 billion transactions. Every single one of those transactions must reach one of exactly two states: completed or rejected. There is no "partially transferred." There is no "mostly settled." A wire instruction moving $400 million between correspondent banks either executes atomically or it does not execute at all. The entire architecture of modern financial infrastructure is built on this binary: success or defined failure, with nothing in between.
Shubhankar Shilpi has spent his career ensuring that binary holds. As an Engineering Director in Banking Technology, he has led the modernization of ACH, Wire, SWIFT, and Treasury systems at Truist -- the institution formed from the merger of BB&T and SunTrust, now the seventh-largest commercial bank in the United States. His patent-holding work on payment processing architecture and his recognition as an innovator in banking technology stem from a single operational truth: financial systems cannot afford ambiguous failure states. When they fail, they must fail in ways that are predictable, traceable, and recoverable.
System Collapse 2026, organized by Hackathon Raptors, inverted that premise entirely. Twenty-six teams spent 72 hours building software where collapse is the feature -- systems that break, mutate, degrade, and adapt as core mechanics rather than defects. Shilpi evaluated eight of those submissions in the event's third batch, applying the same analytical rigor he uses for payment settlement architecture to projects that deliberately embrace the instability banking systems are designed to eliminate.
"In payment processing, we design for exactly two outcomes: the transaction completes correctly, or it rolls back completely," Shilpi explains. "ACID properties are not optional -- they are regulatory requirements enforced by the Federal Reserve, NACHA, and SWIFT. These hackathon projects forced me to think about what happens when you strip away that guarantee. What does software look like when failure isn't a defect to be prevented but a state to be inhabited?"
After the Stroke: Disaster Recovery as a Human Experience
The strongest submission in Shilpi's batch was After the Stroke by team Gladiators, scoring 3.70/5.00 -- the highest in a batch that averaged 2.9 in Technical Execution. The project simulates post-stroke rehabilitation, placing users in an environment where cognitive and motor capabilities have been disrupted and must be rebuilt through sustained effort. The collapse is neurological rather than mechanical, and recovery is nonlinear.
In banking, disaster recovery follows a rigid taxonomy. Recovery Time Objective defines how quickly a system must be restored. Recovery Point Objective defines how much data loss is acceptable. Business Impact Analysis quantifies the cost of each minute of downtime. These frameworks reduce recovery to measurable parameters: hours, dollars, transaction volumes.
"After the Stroke does something that banking DR plans never do," Shilpi observes. "It captures the experience of recovery, not just the metrics of it. When a wire transfer system goes down, we measure time-to-recovery in minutes and settlement impact in dollars. We never measure the cognitive load on the operations team, the decision fatigue during triage, the compounding stress of knowing that every second of delay affects real counterparties."
The project's rehabilitation mechanics -- where users must relearn basic interactions with degraded capabilities -- parallel what Shilpi sees during major system migrations. When Truist formed from the BB&T and SunTrust merger, two entire banking technology ecosystems had to be reconciled into one. Payment rails, account numbering systems, routing configurations, and settlement processes all required integration while live transactions continued processing. "A bank merger is a stroke at the infrastructure level," he says. "The system was functioning, then it was disrupted, and then it had to relearn how to operate with new pathways, new routing, new constraints. The recovery isn't instantaneous. It's iterative, frustrating, and nonlinear -- exactly what this project simulates."
After the Stroke earned its top position in the batch not through technical sophistication but through conceptual honesty. The project does not pretend that recovery returns the system to its original state. It models recovery as adaptation -- the new normal is different from the old normal, and functionality is rebuilt around the damage rather than erasing it.
Life Simulator: Economic Cycles and the Architecture of Burnout
Life Simulator by VrajC0Dee scored 3.00/5.00 and presented what Shilpi describes as a deceptively complex system beneath a simple interface. The project simulates a shop-management experience where economic pressures -- stress, greed, burnout, and renewal -- drive the system's behavior.
"This project delivers a surprisingly thoughtful and emotionally resonant simulation wrapped in a simple shop-management interface," Shilpi noted during evaluation. "Beneath the surface, the system models stress, greed, burnout, and renewal with clarity and intention."
The economic cycle modeling resonates directly with Shilpi's experience in Treasury management systems. Treasury operations at large banks involve constant balancing of liquidity positions, interest rate exposure, and counterparty risk -- a continuous calibration where overextension in any dimension creates systemic vulnerability. Life Simulator compresses this calibration into a single-player experience where the player's own behavioral tendencies (greed, risk tolerance, work ethic) interact with systemic pressures to produce outcomes the player did not explicitly choose.
"In Treasury, we talk about 'risk appetite' as though it's a fixed parameter," Shilpi explains. "But risk appetite changes under stress. A trader who maintains disciplined position limits during normal markets starts taking outsized positions during drawdowns, trying to recover losses. The system becomes more fragile precisely when it needs to be more conservative. Life Simulator captures this dynamic -- the player's response to pressure becomes the primary driver of system instability."
Basel III capital requirements exist precisely because of this pattern. The 2008 financial crisis demonstrated that financial institutions systematically underestimated their risk exposure during periods of growth, then amplified their exposure during periods of contraction. Regulatory frameworks now mandate countercyclical capital buffers -- reserves that grow during good times and absorb losses during bad times. Life Simulator implements an analogous mechanic: the player's resources fluctuate in response to decisions that feel rational in the moment but accumulate toward collapse.
"The collapse mechanic is the most honest part," Shilpi adds. "The simulation doesn't give the player an easy reset. Burnout has lasting effects. Greedy decisions create structural weaknesses that persist. This mirrors how real economic cycles work -- the consequences of overleveraging don't disappear when the market recovers. They embed themselves in the system's architecture."
System Collapse: Cascade Failure and Systemic Risk Contagion
The project that shares the hackathon's name -- System Collapse by team keystone -- earned 3.00/5.00 by building a system where failure propagates through interconnected components. The mechanic is straightforward: when one subsystem degrades, connected subsystems absorb that degradation and pass it forward, amplifying the original disruption until the entire network is affected.
For someone who has spent years building financial networks where exactly this kind of contagion must be prevented, the project triggers immediate pattern recognition.
"This is Herstatt risk visualized as a game," Shilpi says, referencing the 1974 collapse of Bankhaus Herstatt, a German bank whose failure during the middle of a trading day left counterparties across time zones holding unreceived payments. The incident was so disruptive that it led directly to the creation of the Basel Committee on Banking Supervision and, eventually, the CLS Bank settlement system designed to eliminate settlement risk in foreign exchange transactions.
Modern payment systems use multiple mechanisms to contain cascade failure. Circuit breakers halt trading when volatility exceeds thresholds -- the New York Stock Exchange's Level 1, 2, and 3 market-wide circuit breakers trigger at 7%, 13%, and 20% declines respectively. Real-time gross settlement systems like Fedwire process each payment individually rather than batching them, preventing a failed batch from poisoning healthy transactions. Bilateral and multilateral netting reduce the total volume of interbank obligations, limiting the blast radius when a counterparty defaults.
"System Collapse the project does not implement any of these containment mechanisms, and that's actually what makes it instructive," Shilpi observes. "It shows what happens when there are no circuit breakers, no netting, no isolation. The cascade is uncontained. Watching it propagate gives you visceral understanding of why these safety mechanisms exist. Every banker should play this for five minutes -- they'd never question the cost of settlement infrastructure again."
The project's limitation, in Shilpi's assessment, is that it models only the downward cascade. Real financial contagion involves feedback loops: a failing bank's fire sale of assets depresses prices for other institutions holding similar assets, triggering margin calls that force further sales, which depress prices further. The cascade feeds itself. System Collapse captures the linear propagation but not the reflexive amplification that makes real systemic crises so difficult to arrest.
Conway's Game Variant: Agent-Based Modeling and Emergent Market Behavior
Conway's Game Variant by team lawless (3.00/5.00) adapts the cellular automaton into a modified rule system where the standard birth, survival, and death conditions are altered. The resulting emergent behaviors -- patterns that arise from simple local rules without centralized coordination -- mirror a class of computational models that financial institutions increasingly use.
Agent-based modeling has become a significant tool in financial risk analysis since the 2008 crisis exposed the limitations of aggregate statistical models. Where Value at Risk treats market participants as a statistical population, agent-based models treat each participant as an independent actor with individual rules and feedback loops. The aggregate behavior emerges from individual actions rather than being assumed from population statistics.
"Conway's Game of Life is the ancestor of every agent-based financial model in production today," Shilpi explains. "Simple rules, local interactions, emergent macro behavior. When you modify the rules -- as this project does -- you get phase transitions. Stable configurations become unstable. Oscillators die. New patterns emerge that nobody designed. This is exactly what happens when you change a regulation in financial markets."
The Dodd-Frank Act of 2010, for example, introduced swap execution facilities that changed how derivatives were traded. The rule change was targeted and specific. The emergent behavior was complex and unpredictable: liquidity shifted between venues, bid-ask spreads widened in some instruments and narrowed in others, and new patterns of correlated trading appeared that no single regulator had anticipated.
Conway's Game Variant scored 3.00 because its modifications to the rules, while functional, did not fully exploit the emergent potential. "The foundation is sound," Shilpi says. "What I'd want to see is instrumentation -- the ability to observe population dynamics, measure entropy, track phase transitions. Without measurement, emergence is just animation. With measurement, it's a modeling tool."
The Atomic Simulator: Atomic Operations and Transaction Integrity
The Atomic Simulator by TheExperimentalists (3.00/5.00) simulates atomic-scale interactions -- particle behavior, energy states, bonding dynamics. The name itself triggered an immediate association for Shilpi, though the connection runs deeper than wordplay.
Atomicity in database systems -- the "A" in ACID -- means that a transaction either completes entirely or has no effect. A wire transfer debits one account and credits another as a single atomic operation. If the credit fails, the debit is rolled back. The system never enters a state where money has left one account but not arrived in another.
"The name caught my attention, but the mechanics earned my respect," Shilpi says. "Atomic interactions in physics and atomic transactions in databases share a fundamental property: state transitions are discrete. An electron occupies one energy level or another -- it does not exist between levels. A transaction is committed or rolled back -- it does not partially execute. This project simulates the physical version of that discreteness."
The Atomic Simulator models energy state transitions, particle bonding, and system stability within a simulated environment. When particles interact, the system's total energy changes in quantized steps. Stability is a temporary equilibrium that external perturbation can disrupt.
"In payment processing, we deal with settlement finality," Shilpi explains. "Once a Fedwire payment is settled, it is irrevocable -- there is no mechanism to reverse it except by initiating a new payment in the opposite direction. The Atomic Simulator's energy states have similar finality. Once a bond forms, the energy landscape changes permanently. The system cannot simply return to its previous configuration without inputting additional energy. That irreversibility is what makes both atomic physics and financial settlement interesting -- and dangerous."
The project's score reflects solid conceptual grounding with limited technical ambition. The simulation models fundamental interactions accurately but does not push into the complex emergent territory where atomic-scale chaos produces macro-scale consequences. In financial terms, it models individual transactions well but does not yet model the systemic effects those transactions produce when they interact at scale.
The Weakest Batch and What It Reveals About Honest Assessment
Shilpi's batch -- the third of three -- averaged 2.9 in Technical Execution, the lowest across the event. Two projects, Fading Ink by Error 404 and Raptor Editor by team Bisht, scored 2.00/5.00 across all criteria. The batch contained no submissions scoring above 4.0, while the other two batches featured multiple projects in the 4.5-5.0 range.
In banking, this kind of variation between evaluation cohorts is a recognized and managed phenomenon. Credit risk assessments of corporate borrowers are calibrated across review teams to ensure that a "BBB" rating from one analyst means the same thing as a "BBB" from another. Portfolio reviews are stress-tested for inter-rater reliability. Scoring curves are adjusted for difficulty.
"In loan underwriting, we distinguish between a weak borrower in a strong economy and a weak borrower in a weak economy," Shilpi explains. "The absolute score tells you one thing. The relative context tells you another. After the Stroke's 3.70 in this batch represents genuine technical effort and conceptual depth. It would have scored differently in a batch with stronger competition, but the underlying work is legitimate."
The honest assessment framework matters because inflation -- whether in credit ratings or hackathon scores -- creates systemic risk. Ratings agencies that inflated scores on mortgage-backed securities in 2006-2007 contributed directly to the financial crisis. Score inflation in any evaluation system eventually erodes the system's credibility and utility.
"I could have given higher scores across the board," Shilpi says. "The projects showed effort and creativity. But evaluation integrity requires calibration. A 3.0 means 'meets expectations with room for improvement.' Inflating that to a 4.0 to be encouraging does the participants a disservice -- it tells them they've arrived when they still have distance to travel. In banking, we call that 'extend and pretend.' It delays reckoning rather than preventing it."
Settlement Finality and the Architecture of Acceptance
Across his eight evaluations, Shilpi observed a pattern that maps directly to a foundational concept in payment systems: settlement finality. In financial infrastructure, settlement finality is the legal and operational certainty that a completed transaction cannot be reversed. It is the moment when a payment stops being a promise and becomes a fact. The entire architecture of the global financial system -- correspondent banking, central counterparty clearing, real-time gross settlement -- exists to define, enforce, and protect that moment.
The System Collapse hackathon asked teams to build software where collapse is not a failure but a feature. The strongest projects in Shilpi's batch -- After the Stroke, Life Simulator, System Collapse -- achieved something that the weaker ones did not: they gave collapse finality. When the system broke, it stayed broken in meaningful ways. The damage persisted, shaped future behavior, and could not be trivially undone.
"The projects that treated collapse as reversible -- reset the game, try again, no consequences -- missed the theme," Shilpi observes. "Real collapse has settlement finality. The 2008 crisis is settled. Lehman Brothers is gone. The regulatory landscape changed permanently. You cannot roll back systemic failure. The best projects understood this: collapse is irrevocable, and the system that emerges from it is fundamentally different from the system that existed before."
This is the tension that defines both banking technology and the System Collapse theme. Financial systems invest enormous resources in preventing failure -- redundancy, monitoring, circuit breakers, capital reserves, stress testing, regulatory oversight. But when failure occurs despite all prevention, the system must accept it, absorb it, and reorganize around it. The Federal Reserve's emergency facilities during 2008, the Bank of England's response to the LDI crisis in 2022, the Swiss National Bank's orchestration of the UBS-Credit Suisse merger in 2023 -- each represents a system accepting collapse and channeling it toward controlled resolution.
"These hackathon teams built something that most financial engineers spend careers avoiding: systems where collapse is the normal operating mode," Shilpi concludes. "What surprised me is how much the principles transfer. Containment, finality, adaptation, honest assessment of damage -- these apply whether you're managing a failed wire transfer or a failed game world. The engineering discipline is the same. The difference is that in banking, we call it risk management. In this hackathon, they called it gameplay."
System Collapse 2026 was organized by Hackathon Raptors, a Community Interest Company supporting innovation in software development. The event featured 26 teams competing across 72 hours, building systems designed to thrive on instability. Shubhankar Shilpi served as a judge evaluating projects for technical execution, system design, and creativity and expression.
