Day 1 | HALL 2 - Practical Insights and Best Practices

AI has made building intelligent data applications easier than ever—but in global pharma, ease of implementation does not equal trust. Across regions, the same drug, KPI, or business concept carries different identities, definitions, and regulatory meanings. This session explores why naïve AI adoption can produce technically correct yet semantically unsafe insights.

Using real‑world pharma scenarios—global drug identity resolution, cross‑regional market share, and retroactive data corrections—we show where AI falls short and why modern data engineering is critical. The talk highlights how semantic models, metadata‑driven governance, time‑aware lineage, and federated architectures enable trustworthy AI at scale.

The key message: AI sits on top of semantic infrastructure. As AI commoditizes implementation, data engineering’s role evolves from moving data to preserving enterprise meaning, context, and trust across markets and time.

Most enterprise AI initiatives struggle not because the models are inadequate, but because enterprise data, workflows, and execution systems were never designed for operational AI. This session explores why GenAI initiatives fail to scale beyond pilots, how enterprises must evolve from traditional data platforms to intelligent decision systems, and what operating models are required to operationalize AI successfully. The discussion will also cover the evolving role of GCCs and showcase real-world enterprise AI business outcomes.

Data engineers are the backbone of reliable AI, yet too often their expertise is locked inside repetitive recontextualization, one-off scripts, and ad hoc workflows that don’t scale.
This tech talk explores how agent skills empower data engineers to break that cycle by packaging domain knowledge into portable, reusable building blocks that eliminate the need to rephrase the same logic across every new request. From ingestion to governance, attendees will walk away with practical steps and tools to build skills that give AI agents the context to act autonomously, consistently, and at production scale. So, data engineers can stop repeating themselves and start multiplying their impact.

AI adoption often stalls between vision and execution, leaving many enterprises struggling to realize tangible value despite significant investments. This session explores how to operationalize AI by prioritizing high-impact use cases and embedding them into core workflows—while leveraging the right mix of talent, technology, and processes to drive sustained, measurable business outcomes

Data evolution is no longer limited to migration—it requires intelligent modernization. By re-architecting data platforms with governance, scalability, and AI-readiness at the core, organizations can unlock advanced analytics, ensure trusted data, and accelerate innovation. This approach establishes a resilient foundation for enterprise-wide AI adoption and sustained business value.

How L&T Finance is strengthening its data foundation to support scalable, reliable AI use cases—focusing on data quality, governance, and real-time accessibility.

RisingWave is redefining real-time data processing with its next-gen streaming database architecture. This session explores how organizations can move beyond batch systems to build low-latency, scalable pipelines that power instant insights thereby reducing operational complexity of traditional stream processing stacks

This talk will introduce hybrid search with Apache Doris as a next-generation retrieval solution for generative AI and context engineering. Addressing the semantic confusion problem by combining vector search, full-text search, and SQL to capture the exact-match intent with semantic similarity, resulting in a more accurate and cost-effective solution. Matt will then touch on how the native real-time capability that Apache Doris brings to OLAP can be extended to real-time RAG, helping organizations think about future challenges in this space.

Traditional data lakes often struggle to deliver reliable and actionable insights at scale. This session shares how redBus evolved to a modern lakehouse architecture, built a unified knowledge layer for context and governance, and enabled AI-powered, natural language data access—transforming its data platform into a real-time, decision-centric system.

Traditional data pipelines were designed to move and transform data efficiently—but in today’s AI-driven world, that is no longer enough. This session explores the evolution from DataOps to AI-native pipelines, where data systems are not just enablers but active participants in intelligence generation.

We will dive into how modern architectures integrate real-time data processing, feature engineering, model deployment, and feedback loops into a unified AI pipeline. Attendees will gain insights into designing scalable, reliable, and cost-efficient systems using technologies like Spark, Kafka, and cloud-native platforms.