Time-critical software is no longer a niche concern for trading desks or call centers. It now shows up in ordinary moments: a clinician closing a note before the next patient, a customer checking a balance twice because the screen hesitated, a store associate staring at a payment spinner while a line forms. The economics are shifting with it. The application performance monitoring market is expected to grow from $8.43 billion in 2024 to $19.62 billion by 2030, a sign that “fast enough” is becoming a business requirement, not an engineering preference. Hitesh Jodhavat, a Cloud Performance Architect with 25 years of experience and an esteemed reviewer at the 9th International Conference on Computational Intelligence and Data Science (ICCIDS) 2026, who has built his career inside these high-pressure edges, where systems cannot ask users to be patient. To understand how teams build performance that people actually feel, we turned to Jodhavat.
“Clinicians do not experience latency as a metric, they experience it as lost care minutes. When a screen pauses, the visit gets shorter, not the work. The system should disappear into the workflow, not interrupt it.” That is the lens Jodhavat brings to healthcare systems where time does not stretch. Care cannot wait.
That pressure is measurable inside the EHR itself. For every 8 hours of scheduled patient time, physicians can spend 5.8 hours doing EHR work, and primary care physicians can spend another 2.7 hours in the EHR during personal time. Jodhavat’s work on a clinical AI agent and an EHR redesign targeted that compounding drag directly, with a practical goal: give back roughly 50% of clinician time by removing the small delays that accumulate across logins, chart loads, and repetitive documentation. In one readiness review, the moment that mattered was not a benchmark chart, it was watching a context panel arrive just slowly enough to break a clinician’s sentence. His approach treated that pause as the defect, then worked backward through query paths, caching choices, interaction design, and data retrieval until the workflow stopped bleeding minutes.
If healthcare makes time visible, banking makes time personal. A delay is interpreted as uncertainty, and uncertainty becomes doubt. Trust is timing. In consumer banking, that expectation is now mainstream: 55% of consumers prefer using a mobile banking app, while 22% prefer online banking on a laptop or PC. Jodhavat put it this way: “When customers hit refresh, they are testing trust, not speed. A fast page that fails under load is worse than a slower one that stays consistent. The goal is predictability that feels safe.”
That is why his contribution to performance testing for the Chase.com redesign focused on more than a single response-time target. He wrote and executed tests that validated how the experience behaved under real-user load, across the flows that matter most: checking balances, transferring funds, and confirming activity without jitter or surprising crashes. The work lives in the unglamorous details, how time breaks down across the stack, where retries appear, what happens when traffic spikes, memory released at the right time after usage, and whether an “improved” experience stays coherent outside a demo. In banking, the refresh button is a trust probe, and his performance validation was built to keep that probe from turning into doubt.
Over time, every fast system meets the moment that proves whether it is resilient or fragile, the incident. The important unit is not the meeting, it is the minute. Minutes are money. High-impact outages can cost an average of $2 million per hour, and mean time to detection can be 28 minutes. Jodhavat’s view is blunt: “War rooms expose the gaps between what you monitor and what you need to know. You learn which metrics were comforting instead of useful. The fastest fixes come from teams that already rehearsed how to see time.”
That framing shows up in how he co-led a performance engineering practice delivering diagnostics, tool strategy, and hands-on incident support across engagements. The impact is not one tuning win, it is building a repeatable response: define monitoring that explains time, establish a triage rhythm that narrows root cause quickly, and create post-incident learning that reduces recurrence. His service as a Session Chair at The 2nd International Conference on Artificial Intelligence of Things and Computing (AITC 2025) reflects peer-level trust in that judgment, the kind earned by practitioners who can translate system behavior into operational decisions when the clock is running.
Retail is where software loses the luxury of controlled environments. Networks are uneven, devices roam, and the line does not pause. Stores are production. Digital wallets are anticipated to exceed $25 trillion in transaction value by 2027, up from $13.9 trillion in 2023, which means more transactions are riding on experiences that must complete instantly at the point of service. As Jodhavat describes it: “Checkout is not a page load, it is a moment in a line. Every delay is visible to the customer and the associate. Performance is the ability to keep the flow moving.”
That is why his work supporting Apple retail operations treated Mobile POS connectivity failures as first-order business interruptions, not background IT noise. The job is urgency paired with composure: rapid triage under real constraints, clear technical ownership, and restoring the store workflow quickly when the network is not. His outcomes included resolving the vast majority of tickets within one hour, not as a vanity metric but as a requirement of the retail floor. The deeper lesson is that “performance” in retail is often the ability to recover fast, because the customer experience is happening in real time, in public, with no buffer.
Once checkout works, the customer’s next question becomes invisible but decisive: what happens to the order now. That decision is increasingly the storefront. The global omnichannel retailing market was valued at $7.80 billion in 2023 and is estimated to reach $19.51 billion by 2030, a signal that more commerce journeys will span web, store, and fulfillment paths that must stay consistent. In that world, order routing is where customers feel “fast” or “broken,” even when the UI is polished.
Jodhavat’s work on Target’s multichannel order management modernization centered on building an OMS on an Oracle Commerce platform with security-group access control, role-specific permissions, and configuration that could survive real UAT pressure. The performance risk here is not only latency, it is incorrect orchestration that creates rework, exceptions, and dead ends in the journey. That insistence on rigor also shows up in the external technical reviews that rely on his judgment, including his role on the 2025 4th International Conference on Artificial Intelligence and Education (ICAIE 2025) Technical Committee, where submissions are evaluated for clarity, correctness, and technical substance. His closing perspective is that speed and correctness cannot be separated in order systems: “Omnichannel fails when orders are treated like an afterthought. The customer does not care where the delay lived, only that the promise broke. Reliability is a product feature, and it has to be engineered end to end.”