Hello, comp 42. It's an absolute pleasure to be here. My name is Srin Ana, and I lead engineering efforts in Microsoft's finance organization. Today I'm excited to share a real world transformation story, a journey where legacy systems met modern technology. And where measurable business outcomes were achieved by combining cloud and ai. The problem we tackled wasn't unique. Many enterprises face the same friction with aging systems, disconnected data, raising costs, and compliance complexity. But how we solved with discipline, collaboration, and right. Technical levers is what I want you to walk today. We didn't just modernize, we revolutionized financial systems, delivering results like 20%. 25% perhaps reduction in operational costs, 40% improvement in data accuracy, 50% acceleration in delivery cycles. And we did this while supporting global financial operations that run on tight schedules and intense. Regulatory scrutiny. Let's get into that now. When we started, our systems were stitched together with patches, outdated architectures, and fragile integrations. We had decades old reporting tools running on Reise servers. Teams were constantly firefighting, rather in awaiting. The root of the issue was threefold. Legacy systems, they lacked modularity, had hard coded logic and couldn't scale to new use cases. Integration silos, multiple tools and data warehouses were not talking to each other. Finance team spent hours reconciling numbers. Scalability constraints. Business needs grow fast. Our systems couldn't keep up with the pace. The cost wasn't just technical. It showed up in missed deadlines, poor insights and anxiety. During quarter closes, by spending time reconciling the numbers from one place to another, we knew we had to rethink everything, not just migrate, but reimagine. Our first step was to simplify the environment through platform consolidation. Think of this like cleaning out a cluttered attic. You find five different tools doing similar things, disconnected spreadsheets with conflicting formulas and pipeline running on legacy jobs. We took a four step approach audit. We. An inventory of every financial application, reports and integration mapped to a business value. We identified the systems and underused tools as well by leveraging telemetry. Map. We designed a unified architecture with clear integration points, reducing friction between upstream and downstream systems, consolidate. We eliminated redundant tools, standard data pipelines, and focused on reusable components by consolidating the systems as well. Validate this was the key. Every feature reported or rebuilt was tested with performance, security, and compliance. For example, we had three different separate systems generating headcount reports. We merged them into one platform, reducing the efforts by 60%, and ensuring everyone was working with the same numbers. This laid the groundwork for the next phase, which is cloud migration. Cloud is more than just an infrastructure. It's about agility. We leveraged Azure to build a resilient, scalable financial data platform. The results were tangible, 50% is faster. Time to market forecasting models that used to take months. To deploy now go live in within weeks 35% cost optimization. We eliminated redundant storage, shut down legacy VMs, virtual machines, and scaled compute on demand, 99.9% availability. This was critical. Financial systems must be always on, especially during close and forecast cycles. A quick anecdote, we had a year and close deadline where reporting environments used to take nearly 12 hours to prep After cloud migration, the same setup was ready under 20 minutes. That's the kind of a business agility that drives the trust and credibility across the teams. The cloud gave us not just speed, but reliability, security, and much more which is more resilience as well. With clean architecture and cloud scale, we leaned into AI to bring intelligence into our systems. This wasn't just about fancy models, it was about solving real pain points for finance users. We focused on three care key areas. Predictive analytics. We built forecasting models with 85% accuracy using time series and historical signals. This helped finance leaders anticipate OPEX trends, hiring spikes, and vendor spending, automated processing previously manual workflows. Like invoice matching, report scheduling, and cost reconciliation. Were automated. We cut manual efforts by 75%. Anomaly detection. We used realtime AI models to flag, duplicate payments, cost spikes, and misla classified expenses. 30% better persistent over rule-based checks. This made our financial systems more intelligent, responsive, and secure. Analysts could now focus on insights instead of troubleshooting the data. Let's talk about the hidden superpower in our stack. Metadata. Our platform is metadata driven, meaning every dataset, transformation, and the business rule is described and governed by the metadata. This impacts in self-documenting systems. New team members ramp up faster and maintenance becomes much more predictable. Dynamic data models, instead of changing code, we change metadata. That means faster adaptability. To regulatory or business changes. Governance metadata allows full lineage tracking. If someone changes a report logic, we know exactly where it propagates. We treated metadata not as an afterthought, but as a design principle. This paid off in compliance audits, system reliability, and long-term maintainability. Let's look at what all this deliver in real business terms, report generation. Improved from 24 hours to three hours. Data errors reduced from 12% to 3%. Operational costs came down by 25%. Time to insights changed from 48 hours to 12 hours. These are just not metrics. They reflect fewer escalations, more confidence leadership decisions, and an engineering team that's respected, not firefighting. We even heard from our CFS office that our modernization platform was now considered the source of truth for executive dashboards. That's a huge order of trust in engineering. Technology is only one piece. Success required execution discipline. This comes with strategic planning, every initiative tied to a business goal. Agile development, sprints, stakeholders, feedback with faster iteration change management. We trained users, provided shadowing, and had hands-on labs as well. Continuous improvement Every release had built in feedback loops. We also built strong cross team partnerships between finance, SMEs, subject matter expertise, data engineers, security teams, and product managers. No one worked in a vacuum. This was the key to not just delivering transformation, but sustaining as well. As I mentioned initially, this has been a journey. All journeys comes with learnings as well. Here are some of the biggest lessons that we learned. Things we would do it again, and some we would do it differently. Start small scale fast. Don't boil the ocean, prove value in one area, and then expand cross-functional alignment. Engineers alone can't solve financial systems. Ambit finance experts in the development process as well to expedite it, prioritize data quality. You can have the best AI model in the world, but garbage in and garbage out. These insights help us to build a system that works not just today, but in future Ready as well. To wrap up, I would like to say modernizing financial systems is no longer optional. It's a mission critical. If you are facing similar challenges, I encourage you to start with what's broken and quantify the business plan. Sometimes you would be surprised what kind of damage the existing legacy systems can do to your time, to your data quality, and to the credibility of the systems as well. Don't skip governance. It's the foundational of trust. Invest in the right levers, cloud for scale, AI for insights, metadata flow for flexibility. This journey isn't easy. But the rewards are enormous, faster insights, trusted data and operational excellence. Thank you so much for joining me. I'm happy to take any questions or chat afterward. Let's build the systems that work for the business, not the other way around. Thank you very much.