Abstract is Coming Soon

As Generative AI transforms how organizations create, consume, and act on data, traditional data strategies are being pushed to their limits. In this session, I’ll share how enterprises can rethink their data foundation to support scalable, secure, and trustworthy AI adoption.
Drawing from real-world experience leading supply chain analytics and data product strategy at H-E-B, this talk covers:
• Key pillars of a modern data strategy tailored for Gen AI
• How to bridge the gap between legacy data systems and AI readiness
• Data governance, quality, and privacy frameworks in the Gen AI context
• Organizational alignment: turning data chaos into AI enablement

Abstract is Coming Soon

Many companies pursue product or performance improvements without fully understanding how those changes impact user behavior or business value. This talk explores how organizations can apply data science and experimentation to more effectively connect digital experience enhancements—on websites or apps—to top-line outcomes like revenue, engagement, and retention. We’ll discuss how to frame performance changes as testable hypotheses, run targeted experiments, and analyze results to uncover the causal relationships between operational metrics and business impact. A key focus will be on developing and using transfer functions to quantify how improvements in key performance indicators translate into enterprise value. Attendees will learn how to use these techniques not just for one-off experiments, but as a scalable framework for continuous optimization and goal tracking across teams.

This talk delves into the transformative power of causal inference and experimentation in operations, emphasizing how rigorous experimentation can dramatically enhance operational decision-making and strategic outcomes. By integrating principles from causal inference—such as randomized controlled trials (RCTs), natural experiments, and observational data analysis—organizations can more accurately determine cause-and-effect relationships within complex operational processes.
Attendees will learn practical methodologies for designing effective experiments, interpreting results, and avoiding common pitfalls such as confounding variables, selection bias, and reverse causation. The discussion will include real-world case studies demonstrating successful applications of causal inference in workforce productivity, key operation metrics and customer sentiments.

As data science becomes more democratized, ensuring efficient, scalable, and reproducible ML operations is critical. This talk introduces DataOps—a methodology that combines DevOps principles with data science to streamline the model development lifecycle. We’ll explore CI/CD pipelines for ML, automated data versioning, feature stores, and monitoring strategies that help organizations manage data and models at scale. By the end of this session, attendees will gain insights into how to implement DataOps to support democratized data science workflows while maintaining governance and security.

Financial AI systems for loan approvals, fraud detection, and credit scoring lack the transparency needed for regulatory compliance and bias detection, as current systems prioritize performance over auditability. We introduce an auditable AI Framework with four integrated layers: Data Provenance Logging (tracks data origin and transformations), Model Lineage Tracking (documents training artifacts and configurations), Decision Traceability Logs (stores inference metadata and SHAP explanations), and a Compliance Rules Engine (validates outcomes against anti-discrimination policies). Our case study demonstrates the framework through an AI loan decision system using synthetic datasets to simulate regulatory audits, integrating MLflow, SHAP, and Fairlearn, enabling automated audit log generation and a governance dashboard providing risk teams with human-readable compliance summaries. This architecture embeds auditability as a core capability rather than an afterthought, enabling financial institutions to meet evolving AI governance requirements (EU AI Act, ECOA, Dodd-Frank) while maintaining model performance and bridging the gap between AI intelligence and regulatory accountability.

Healthcare organizations are facing unprecedented challenges in managing exponentially growing patient data, with the average hospital now generating over 50 petabytes of data annually. Traditional monolithic architectures have proven inadequate, with 76% of healthcare CIOs reporting scalability issues and 68% experiencing system downtime that directly impacts patient care.
This presentation explores how distributed systems are revolutionizing healthcare delivery by addressing these critical challenges. With microservices architecture adoption in healthcare growing at 24% annually since 2020, the industry is witnessing a fundamental shift in how digital health platforms are designed and deployed.
Our analysis of 150+ healthcare institutions that transitioned to distributed systems reveals compelling outcomes: 87% reported improved system uptime (from 99.1% to 99.97%), 63% reduced infrastructure costs through targeted scaling, and 92% experienced faster innovation cycles—reducing new feature deployment time from months to days.
We’ll examine how horizontal scaling enables healthcare platforms to accommodate 30-40% annual increases in patient data volume without performance degradation. Case studies will demonstrate how fault tolerance mechanisms have reduced critical system downtime by 94%, ensuring continuous access to patient records even during regional outages or cyberattacks.
Additionally, we’ll explore how load-balancing techniques have improved response times for critical healthcare applications by 72%, with appointment booking systems handling 5x more concurrent users during peak periods without degradation.
For healthcare technology leaders navigating this transition, this session provides a roadmap to implementing distributed architectures that enhance scalability, reliability, and interoperability—ultimately delivering superior patient experiences across the care continuum.

Large language models can now power capable software agents, yet real‑world success comes from disciplined engineering rather than flashy frameworks. Most reliable agents are built from simple, composable patterns instead of heavy abstractions.
The talk will introduce several patterns that add complexity / autonomy only when it pays off:
1. Augmented LLM (retrieval, tools, memory) as the atomic building block
2. Workflow motifs: prompt chaining, routing, parallelization etc with concrete criteria and implementation tips
3. Autonomous agents that loop through plan‑act‑observe‑reflect cycles to tackle open‑ended tasks
Attendees will leave with a practical decision framework for escalating from a single prompt to a multi‑step agent, reference implementations they can reproduce in a few lines of code, and robust guardrails for shipping trustworthy, cost‑effective agents at scale.