Data Science stack

Data Science podcast

Ten long-form episodes per stack: people who shipped real Data Science software for messy human problems—inventory, bookings, tours, money, and the spreadsheets that came before. Each page has details, show notes, timestamps, a YouTube link, and a full transcript.

Data Science episodes

6 episodes

Why Some Data Science Architectures Survive: Boundaries, Testing, and Maintainability in Real Teams

#1: Why Some Data Science Architectures Survive: Boundaries, Testing, and Maintainability in Real Teams

Why Some Data Science Architectures Survive: Boundaries, Testing, and Maintainability in Real Teams

In this episode, we dig deep into the architecture patterns that help data science projects endure the realities of real-world teams and production environments. Our guest shares first-hand experiences building, breaking, and rebuilding data workflows, emphasizing the critical roles of clear boundaries, robust testing, and maintainability. Listeners will gain practical strategies to overcome the messiness of handoffs, misunderstandings, and legacy code, and learn how thoughtful design can future-proof data products. We also explore how team dynamics and communication styles influence architectural choices, and discuss patterns that can help avoid common pitfalls in scaling, refactoring, and operationalizing models. Whether you’re a data scientist, ML engineer, or technical lead, this conversation offers actionable insights for building systems that last.

View Episode
Profiling, Bottlenecks, and Optimizing Data Science Workflows: A Real-World Deep Dive

#2: Profiling, Bottlenecks, and Optimizing Data Science Workflows: A Real-World Deep Dive

Profiling, Bottlenecks, and Optimizing Data Science Workflows: A Real-World Deep Dive

In this episode, we take listeners on a hands-on exploration of data science performance: from identifying bottlenecks to making practical, impactful optimizations. Our guest walks through the modern tools and strategies for profiling code, diagnosing slow pipelines, and balancing resource use in both research and production environments. We tackle the real challenges teams face—from misleading metrics and memory leaks to the hidden costs of data loading and feature engineering. Along the way, we share anonymized mini case studies, unpack common pitfalls, and debate when to optimize versus when to refactor. Whether you’re a data scientist, machine learning engineer, or analytics lead, you’ll gain actionable insights to improve your team’s workflows and deliver faster, more reliable results.

View Episode
Building Robust Data Science APIs: Idempotency, Rate Limits, and Failure Modes

#3: Building Robust Data Science APIs: Idempotency, Rate Limits, and Failure Modes

Building Robust Data Science APIs: Idempotency, Rate Limits, and Failure Modes

What happens when data science systems meet the unpredictable world of APIs and integrations? In this episode, we explore how modern teams build resilient, scalable interfaces that can handle everything from duplicate requests to system overloads and real-world failures. Our guest shares practical stories of where integrations went sideways, why idempotency is non-negotiable, and how rate limiting can shape user and model behavior. You’ll learn how to anticipate API edge cases, design safe retry strategies, and avoid the hidden traps that derail production data science. Whether you’re architecting new pipelines or making legacy systems safer, this conversation is packed with actionable advice and war stories from the field.

View Episode
Security Pitfalls in Data Science Apps: Auth, Secrets, Supply Chain, and Safer Defaults

#4: Security Pitfalls in Data Science Apps: Auth, Secrets, Supply Chain, and Safer Defaults

Security Pitfalls in Data Science Apps: Auth, Secrets, Supply Chain, and Safer Defaults

Data science applications are powering more business decisions than ever, but many teams overlook critical security pitfalls in their rush to deploy. In this episode, we dive deep into the unseen vulnerabilities that can undermine your models and data products. We break down practical challenges around authentication, the mishandling of secrets, dependency risks in the supply chain, and the often-overlooked dangers of insecure default configurations. Along the way, we share anonymized stories of high-impact incidents, debate best practices, and uncover why common 'quick fixes' can backfire. Whether you're a data scientist, engineer, or leader, you'll come away with actionable strategies to make your data science apps safer, more resilient, and ready for real-world scrutiny.

View Episode
Operational Excellence in Data Science: Monitoring, Incident Response, and Deployment Discipline

#5: Operational Excellence in Data Science: Monitoring, Incident Response, and Deployment Discipline

Operational Excellence in Data Science: Monitoring, Incident Response, and Deployment Discipline

What does it take for data science products to thrive in the relentless demands of production? In this episode, we dig into the operational backbone of successful data science teams: robust monitoring, effective incident response, and disciplined deployment practices. Our guest shares real-world stories about what happens when data products go sideways, how to design systems for observability, and why deployment discipline is crucial to avoid midnight fire drills. Listeners will gain actionable insights into building feedback loops, the human factors of incident response, and the art of balancing agility with reliability. Expect practical frameworks, lessons learned from production failures, and perspectives on scaling operational maturity as data science becomes increasingly mission-critical.

View Episode
Future-Proof Data Modeling and Migrations: Strategies to Avoid Painful Rewrites

#6: Future-Proof Data Modeling and Migrations: Strategies to Avoid Painful Rewrites

Future-Proof Data Modeling and Migrations: Strategies to Avoid Painful Rewrites

Designing robust data models and planning seamless migrations are among the most daunting challenges in data science projects. In this episode, we delve into actionable techniques for architecting data models that can evolve, how to anticipate and mitigate the risk of disruptive rewrites, and what real teams do to keep their production systems healthy as requirements change. Our guest shares hard-won lessons from leading high-impact migrations and exposes common traps that cause data science efforts to grind to a halt. We’ll uncover the subtle signals that a rewrite is looming, how to build migration playbooks, and why the right abstraction layers can mean the difference between a weekend upgrade and months of fire-fighting. Whether you’re wrangling tabular data, deep learning features, or streaming pipelines, this conversation will help you put the foundations in place to scale and adapt. If you’ve ever groaned at a brittle schema or lost sleep over a failed migration, this episode is for you.

View Episode

More Episodes by Stack