For two Chicago product teams, the stakes are high for product releases.
Technologists at NinjaTrader and Pangea Money Transfer have to balance their product releases with the need for speed and the necessary precision that comes with working in fintech. A misstep can have a significant impact on clients. Luckily, NinjaTrader CTO Adi Nishandar has a strategy.
“Speed is a byproduct of safety, not a tradeoff against it,” Nishandar said. “That’s our rule.”
Nishandar shared that the volume of global customers and around-the-clock financial movement requires the team to approach their craft with care.
“When you operate at that scale, a bad release doesn’t just frustrate users,” Nishandar said. “It can move real money in the wrong direction, trigger regulatory scrutiny and erode the trust that traders place in us every time they put capital at risk.”
The engineering team at Pangea Money Transfer takes a similarly careful approach — building out product changes one brick at a time.
“We ship small, easily reversible changes that are simple to attribute issues to, with active and passive monitoring in place,” Engineering Manager Chris Nielson said.
Built In spoke with Nielson and Nishandar to learn how the two fintech product teams build with care, precision and speed.
NinjaTrader Group is a fintech company that provides software and brokerage services to help traders work smarter.
What’s your rule for fast, safe releases — and what KPI proves it works?
Speed is a byproduct of safety, not a tradeoff against it. That’s our rule.
NinjaTrader serves over 2.5 million customers globally in a low-latency trading environment that runs 24/7. When you operate at that scale, a bad release doesn’t just frustrate users. It can move real money in the wrong direction, trigger regulatory scrutiny and erode the trust that traders place in us every time they put capital at risk. Shipping recklessly is irresponsible.
So, we’ve built a release culture where guardrails actually accelerate us rather than slow us down. Every release passes through progressive exposure gates. We use feature flags to control rollout and easy rollbacks. Real-time telemetry monitors the operational metrics that actually matter for a trading business, like order execution accuracy and latency.
What’s it like to work on the engineering team at NinjaTrader?
“We’ve built a release culture where guardrails actually accelerate us rather than slow us down. Every release passes through progressive exposure gates. We use feature flags to control rollout and easy rollbacks.”
—Adi Nishandar, CTO
We also enforce what I call a “no surprises” principle. If a change touches the order lifecycle or risk systems, it goes through a heavier review gate, with our chief architect and other senior engineers who understand the complexity of the trading domain and the distributed systems.
What standard or metric defines “quality” in your stack?
Quality in our stack comes down to two metrics that work together: bug escape rate and bug pickup SLO.
Bug escape rate tells us how many defects make it past our testing and review gates into production. This is the upstream signal. It measures how good our process is at catching problems before customers see them. When this number spikes, it means something in our pipeline needs attention, whether that’s test coverage, code review rigor or the complexity of what we’re shipping.
Bug pickup SLO is the downstream signal. Once a bug does escape, how fast do we respond? Our standard is that 90 percent of bug PRs must be picked up within 24 hours. In a 24/7 trading environment, a bug sitting in a queue isn’t just a backlog item. It’s a risk that compounds with every trading session.
The combination matters. A low escape rate with a slow pickup time means you’re good at prevention but bad at response. A fast pickup time with a high escape rate means you’re just really efficient at fighting fires. You need both.
We track these at the team level, which is where the data gets actionable.
Name one AI/automation that shipped recently and its impact on your team or the business.
We recently rolled out Cursor and Claude Code across our engineering organization and the impact showed up faster than I expected.
The biggest win has been production debugging. Our trading infrastructure is complex — distributed systems, real-time data flows and multiple execution venues. When something breaks in production, the diagnosis often requires an engineer to hold a lot of context in their head at once. We built Claude Skills, reusable prompt templates that encode our system knowledge and debugging patterns. Engineers point Claude at logs, stack traces and the relevant code paths and get to root cause significantly faster. Our mean time to resolution has improved meaningfully. It doesn’t replace the engineer’s judgment, but it compresses the investigation cycle. In a 24/7 trading environment, that time compression directly reduces customer impact.
Pangea Money Transfer is a digital money transfer platform that makes sending money abroad simple, reliable, and cost-effective.
What’s your rule for fast, safe releases — and what KPI proves it works?
We ship small, easily reversible changes that are simple to attribute issues to, with active and passive monitoring in place.
We validate releases by monitoring system health and customer signals — including latency, server utilization, support volume and business metrics like cancellation and conversion rates. By comparing trends and outcomes between test and control groups, we can measure real business impact while keeping production risk low.
What’s it like to work on a project on the engineering team at Pangea Money Transfer?
“We validate releases by monitoring system health and customer signals… By comparing trends and outcomes between test and control groups, we can measure real business impact while keeping production risk low.”
—Chris Nielson, Engineering Manager
What standard or metric defines “quality” in your stack?
Our primary quality metric is “customers impacted.” While signals like a latency spike from 75ms to 150ms are worth investigating, our true focus is ensuring reliability for customers. We post-mortem every production incident and quantify customer impact, using that data to identify and eliminate problematic trends over time.
Name one AI/automation that shipped recently and its impact on your team or the business.
We recently developed a suite of internal tools and AI-agent-focused documentation to help transition our teams toward AI-first development. The shift has required rethinking how we work and we’re increasingly acting as orchestrators of these tools while remaining accountable for outcomes.
The result has been measurable gains in developer productivity and faster iteration on internal tooling, allowing teams to spend more time on higher-impact engineering problems.
