Hello everyone. Good morning or good evening, depending on your location. My name is MuTu and I have 23 years of experience across data engineering, AI and ML Cloud technologies, enterprise, BI data science, and in the manufacturing sector. It's pleasure to be here at Con 42 to share insight on the topic. AI driven predictive maintenance that transforming the backbone of manufacturing. For decades, factories have relayed on two extremes. First one is waiting until the occupant fails and then fix it, which is reactive maintenance or carrying out scheduled preventive maintenance that often wasting valuable resource. But today we stand at pivotal moment. With the power of AI data and connector systems, we can move beyond reactive. Fire factoring and embrace proactive intelligence maintenance that drives efficiency and resilience. Let's get started. Here is the agenda today. First, critical inflection point in the manufacturing. Second, AI driven maintenance resolution. Third, evolution of maintenance strategies. For the technological Archite architecture, enabling transformation, fifth advanced AI methodology in industrial environments. Six Real world stories. Seven. Why 60% of predictive maintenance initiative fail? Eight. Implementation of blueprint for success. Nine. Emerging technology and the future trends. 10. Actionable roadmap for implementation. 11 key takeaway. Let's start with industry challenges, critical inflection point in manufacturing. Let's see why AI based approach is essential. Let's look at the problem and the high cost of reactive maintenance. Manufacturers today are under the pressure and a traditional way of maintaining equipment are simply too costly to sustain reactive maintenance. When equipment fails without warning, the impact is massive up to three times. Three to 10 times more expensive than predictive maintenance. The average loss during unplanned downtime is up to $50,000 per hour. This includes production loss, labor inefficiency, and sometimes even penalties for late shipments. The market speaks for itself. The AI in maintenance sector is expected to grow from 4 billion to nearly 16 billion by 2028. These numbers highlights why manufacturers cannot afford to stay in reactive maintenance. Moving on to next AI driven maintenance solution. Let's see what happens when companies embrace AI for maintenance. One first industry implementing. AI maintenance have reported 50% reduction in downtime, which is Hal the mission, offline time compared to before. Second, 32% decrease in overall maintenance cost, which includes parts, labor, and energy savings. Third, perhaps the most impressive is 385% return on investment. These project don't just play themselves. They multiply value delivered fourth, the models provides provide eight to 12 days. Advanced warning, which is critical for planning downtime and avoiding production disruptions. This means organizations can plan interventions, orders, spare parts ahead of time, and prevent crisis before they happen. This represents shifting from reactive maintenance to strategic ascent management where decisions are data driven. Next evolution of maintenance strategy. Let's take a step back and look at how maintenance has evolved. First, reactive maintenance equipment is fixed only after it breaks, which is leading to emergency fix. Also, it costs us unpredictable cost and a major production disruption. Second Preventive maintenance here. Maintenance follows Fixed schedule, often replacing pos too early or too late here. Downtime is reduced, but inefficiency is there, and also waste of resource also, is there third Predictive maintenance. It is a conditional based maintenance and a service. The equipment as needed basis. This method minimize disruption while keeping the cost under control. Fourth, AI driven predictive excellence here. Machine learnings predicts failure up to 88%. Accuracy also provides eight to 12 days advance warning and continuous improve over time. Here we are moving from rigid schedule two dynamic and data driven decisions, moving to the next technological architectures, enabling transformations. What make this possible? Let's talk about technology stack Number one, IOT sensor network. Industrial missions Accu equipped with sensor with the capturing real time data like temperature, vibration, pressure, et cetera. Each setup generates 1.5 to 2.3 terabyte of daily data, and also enables continuous health monitoring. Second edge computing. Onsite processing cuts response time by 75 to 80 per 85% and it pre-filter sensor data. It also supports instant alert and rapid decision making for critical equipment conditions. Third cloud data platform. It is a scalable infrastructure stores historical data and powers. Powers Advanced ML training and deployment. It enables predictive model and cross facility performance benchmarking. All together these layers gives us both speed and intelligence. Moving on to the next advanced AI methodology. First auto encoder based anomaly deduction. It learns normal equipment pattern to spot deviations. It also, it achieves 90% accuracy with 31% fewer false alarms. Second random forecast forest classifier. It classifies and pinpoints specific failure mode. It also delivers 78 8% precision in the predictive diagnostics. Third recurrent neural network IT analysis analyze time serious sensor data over long period. Also, it deducts degradation pattern before it failures. Fourth digital twin simulations. It creates virtual replica of physical equipment. Also, it boost prediction accuracy by 67%. Throw simulation insights. Each technique has a role depending on the context. Next, let's see the real world success stories. Why 60% of predictive maintenance initiative fails. Of course, not every initiative succeeds. In fact, about 60% of project fails. Let's see why it fails. One, data ingestion complexities. Nearly 80% of deployment struggles with the legacy machine, siloed IT system, and inconsistent data quality. Also, lack of sensor and poor standardization. Slow down, reliable predictive insights. Second scope and budget management project often exceed budget by. 30 to 40% due to underestimated complexity. Overly wide scope, weak proof of concept validation leads to cost, overturn, and delayed return on investment. Third, organizational resistance. Around 65% of technicians resistant, resist, adopt adoption because fear in the job loss or not trusting AI recommendations in inefficient training and weak change management amplify resistance to new system. Predictive maintenance is not just a technology problem, it is also about people and the process. Moving to the next implementation blueprint for success, first, value, first approach. We need to start with the most critical equipment where the failure. Cost is most and also focus on high impact wins before wider rollout. Second, strategic data integration build. Yeah, unified data architecture by linking maintenance, production and quality system breakdown silos. To enable seamless predict two insights. Third, human center design. It engage technicians early in the process to shape tools and workflows. It creates. Natural interface that builds trust and adoptions. Fourth phase to scaling. We need to scale step by step with the return on investment validation at each stage, and also prevent budget over time while ensuring measurable business value. When you follow this blueprint, adoption improves significantly. Emerging technology and the future trends. Let's see what is next here. First. Explainable ai. It uses transparent algorithm that makes predictions human interpretable. It builds trust among technicians and also increase adoption by 47%. Second system level monitor, it goes beyond single component to capture, cascading effect. It enables entire insight across the Interconnector production system. Third, sustainability applications. AI driven maintenance cuts unnecessarily energy and resource U usage. It lowers emission by 500 to 1500 metric tons per facility each year. This is about building a smart, sustainable factories X number roadmap. Here is the practical roadmap. First is foundation. It takes one to three months time to build a foundation, access equipment, criticality, set. Baseline metrics and deploy sensor on top assets. Begin data collection and integration to build the foundation for predictive insights. Second pilot implementations. This process takes four to six months of time. Apply ML models to highest impact equipment and then train the maintenance team on the new workflows. Track KPIs, validate re results, and document early return on investment to build the momentum. Third expansion. It takes seven to 12 months of time scale deployment. To the secondary assets and refine models from the pilot learning. Integrate with ERP production planning and rollout. Structured change management. Fourth in the second year. Work on the optimization. Optimization and innovation. Leverage advanced analytics under digital twin for higher accuracy, develop custom ml and establish a predictive maintenance center of excellence. Key takeaway to wrap up here are the key takeaway. One proven return on investment, AI driven maintenance, cut, cut down cost by 50%, and also lower cost by 32%, and also delivers 385% return on investment in three years. Second, strategic implementations success come from starting from high impact equipment, unifying data system, and engaging maintenance team early. Third future proof operations. The next step is moving from isolated predictive tool to fully integrated smart manufacturing ecosystem. Thank you so much for your time. I'm happy to take any questions. Bye. See you later.