Mirko Peters
139 episodes listed below
Enterprise AI, measured like a business.
DataScience Show Podcast is a business-focused AI and analytics series built around practical executive frameworks. Host Mirko Peters treats models, data products, features, and incidents as operating decisions with owners, metrics, budgets, and governance gates.
PRESS PLAY
DataScience Show Podcast treats enterprise AI as a management discipline. The center of the show is not novelty; it is operational control. Host Mirko Peters repeatedly returns to the same hard question: how does a model become measurable business value? The answer usually comes as a playbook. Episodes lay out gates, roles, templates, KPIs, review cycles, and decision rights for senior teams trying to move beyond experiments. AI investment portfolios are scored by value, feasibility, risk, and strategic fit. Funding is released in tranches. Capacity planning turns data, talent, compute, and governance into defensible multi-year forecasts rather than vague ambition. Data contracts become small but enforceable agreements with owners, SLOs, observability, runbooks, and change policies. Feature engineering is treated as a product problem, where reusable signals need lineage, versioning, freshness targets, and consumer contracts. The show is especially strong on the messy middle between technical promise and executive responsibility. It explains why a pilot can prove the technology and still fail commercially. It also explains why a model failure is not just an engineering ticket when underwriting, pricing, supply chains, or customer safety are affected. Incident-response episodes push for predefined thresholds, rollback rules, forensic snapshots, legal coordination, and board-level escalation paths. Human-in-the-loop automation gets the same treatment. Confidence bands, override rates, cost per decision, latency, and business impact replace vague claims about keeping humans involved. The tone is instructional and spare. It works best for listeners who already understand the vocabulary of data platforms and want a cleaner operating model for decisions, budgets, and accountability. The show can feel repetitive and occasionally interrupted by ads or rough transcript artifacts. Still, its value is clear. It turns AI governance from an abstract concern into a set of practical executive moves.
Made for: The show suits C-level leaders, data officers, technology leaders, heads of analytics, product leaders, and data practitioners responsible for scaling AI inside organizations. It is especially useful for people who need frameworks for funding, governance, incident response, and production readiness.
What sets it apart: DataScience Show Podcast stands apart by treating AI strategy as a portfolio, service, and product-management problem rather than a stream of trend commentary. Its episodes translate technical concepts into executive controls: gates, SLAs, ownership, metrics, runbooks, and budget logic.
Welcome to The DataScience Show, hosted by Mirko Peters — your daily source for everything data! Every weekday, Mirko delivers fresh insights into the exciting world of data science, artificial intelligence (AI), machine learning (ML), big data, and advanced analytics. Whether you’re new to the field or an experienced data professional, you’ll get expert interviews, real-world case studies, AI breakthroughs, tech trends, and practical career tips to keep you ahead of the curve. Mirko explores how data is reshaping industries like finance, healthcare, marketing, and technology, providing actionable knowledge you can use right away. Stay updated on the latest tools...
Based on 6 episodes we listened to · September 2026
Mirko Peters turns AI and data strategy into practical questions about governance, communication, adoption, and measurable business value.
Mirko Peters treats data work as an operating discipline, connecting technical choices to decisions about investment, accountability, and measurable business value.
The AI aha moment arrives when your maintenance records fail the data-quality test.
Press play if you want
Short reviews from the PlayNext desk, based on the episodes we processed.
A grounded case for explicit grain, stable keys, and model design that keeps pipelines out of firefighting mode.
Data modeling gets treated here as a reliability discipline, not just a warehouse design choice. When row meaning is fuzzy, KPIs drift and conversion rates change depending on the analyst query.
A systems-first case for letting AI handle more than one task at a time.
The piece treats AI entrepreneurship as a systems problem, and that framing gives it real clarity. It connects one setup to research content ideas, draft social posts, generate ads, and prepare customer replies, then carries that logic through acquisition and onboarding.
Data lineage is framed as the line between AI value and liability.
The episode makes a clear case that AI value and AI liability often hinge on data governance, especially lineage, not on model spectacle. It defines lineage as the documented history of data from source to final use, then shows how easy it is for organizations to invest heavily…
Brand-specific AI is most effective when proprietary data, structured prompts, and human review work together.
The piece makes a clear case that AI is useful for marketing only when it draws on proprietary data and brand-specific context. Its main strength is the practical sequence it presents: structured prompts, expert review, and iterative refinement, followed by evidence of…
AI value starts with the data foundation, then the business problem, then the model.
The piece frames AI as a business choice, not a technical flourish, and keeps attention on whether the data can actually support the use case.
A practical case for data leadership as the work of turning models into decisions.
The episode makes a focused case that data leadership is the bridge between gathering information and changing how a business actually decides.
A clear, practical argument for turning fragmented sensor streams into IoT products customers can value.
This single-episode review finds its hook in a clear modern problem: car sensor data may be valuable, but it is often locked behind fragmentation across sensors and formats.
A clear framework for deciding when a dashboard should stay focused on monitoring.
This episode lays out a useful way to choose the right dashboard approach, and it is strongest when the job is recurring monitoring rather than broad storytelling.
Markets, macro, AI, and crypto in motion.
This episode explores how C-suite leaders can transform fractured feature engineering into repeatable, fundable ML assets to reduce duplication and improve business impact.
This episode provides a practical playbook for C-suite leaders to stress test machine learning systems, implement observability measures, and establish recovery protocols to mitigate business risks from model failures.
This episode provides a decision-first playbook for transforming informal data handoffs into funded, board-ready services with an SLA template, costing model, and 90-day pilot to ensure predictable, trustworthy data.
This episode presents an executive playbook for certifying, procuring, and trusting synthetic training data, using real-world vignettes to illustrate the risks of improper governance.
This episode presents a C-suite playbook called 'uncertainty accounting' to measure, budget, and hedge against overconfidence in AI and data models to prevent costly business failures.
Modeling clarity comes first.
The episode opens with a 9 a.m.
The host opens with a direct question about what separates AI that creates business value from AI that creates business liability.
The host opens with a straightforward complaint.
The episode builds toward a practical AI aha moment for business leaders.
This episode introduces a conductor-like approach to data management, using orchestration and visualization techniques to address the high failure rate of data projects by improving coordination across teams.
This DataScience Show Podcast episode presents the Seven Rings of Data Leadership — a framework developed from interviews with over 200 data leaders — arguing that over 85% of data initiatives fail not from technical shortcomings but from a leadership gap between technical outputs and business outcomes.
This DataScience Show Podcast episode delivers a structured tutorial on transforming raw IoT sensor data into commercial products, using the 'data as oil' analogy as its central frame.
The episode draws a line between dashboards and other forms of data communication.