AI hype, measured against mechanisms.
Data Science at Home treats AI and data science as engineering, economics, and power rather than spectacle. Recent episodes move from recommendation feedback loops and spyware risk to LLM serving, enterprise ROI, programmable money, and energy consumption.
PRESS PLAY
Data Science at Home is a skeptical technology show about AI, data science, and the systems that grow around them. Its recent run is especially interested in what happens after the demo works. The show asks who pays for the compute, where the data goes, which incentives shape the feed, and what kind of control disappears when convenience becomes infrastructure. Social media is treated as a feedback system, not just a cultural annoyance. Recommendation engines become a way to talk about preferential attachment, engagement signals, and attention extraction. Privacy episodes move with the same technical pressure. Device security is examined through spyware, locked bootloaders, auditability, and the uncomfortable tradeoffs of trusting a single company to patch faster than attackers. AI episodes resist the fantasy that large language models are minds. They describe model weights, GPUs, serving architecture, quantization, energy use, and the operational realities behind chat interfaces that look effortless from the outside. The show also keeps one foot in business reality. Conversations with Matt Lee and Josh Miramant focus on AI execution, ROI math, experiments, data quality, governance, production systems, and the discipline required before an expensive tool becomes useful. That practical streak gives the criticism weight. It is not anti-technology. It is hostile to vague claims, unchecked spending, fragile architecture, and platforms that confuse extraction with value. Even when the subject shifts to programmable money, the underlying question remains consistent: who owns the system, who can inspect it, and who is left without a fallback when it fails. The pacing is measured and argumentative. Technical terms appear often, but usually with a plain example close behind. The show suits listeners who want AI and data science explained through mechanisms, incentives, and consequences rather than product announcements.
Made for: For technically curious listeners, engineers, data practitioners, security-minded professionals, and skeptical business leaders who want AI explained through infrastructure, incentives, and operational tradeoffs.
What sets it apart: The show connects low-level technical details to social and business consequences, moving from GPUs, bootloaders, and data pipelines to attention markets, privacy failure, and ROI discipline.
Cutting through AI bullsh*t
Based on 5 episodes we listened to · September 2026
Data Science at Home makes AI and platform tradeoffs legible by following technical choices through production, cost, security, and social consequences.
Data Science at Home keeps asking what happens after a technical idea meets production costs, organizational constraints, or consequences in the wider world.
Francesco asks for sales numbers before your shiny AI idea gets an engineering team.
Press play if you want
Short reviews from the PlayNext desk, based on the episodes we processed.
A sharp account of how feeds reward runaway amplification, not value.
The episode presents social media as a feedback loop, using a storm-scarred forest to show how the tree with the most sunlight and water keeps pulling ahead while smaller ones fade.
Apple’s security pitch runs into the realities of attack surface and vendor dependence.
The piece frames smartphone privacy as an engineering tradeoff, not a guarantee. It argues that Apple’s locked bootloader, App Store-only software, and boot-to-runtime verification can raise security, while also leaving the device reliant on Apple to spot and patch flaws before…
A budget-first AI conversation that keeps asking whether the feature really pays for itself.
Francesco and Matt Lee keep the episode grounded in a practical question: how AI proves its value once it moves past hype and into production spending.
A grounded look at AI moving into production, with governance, testing, and monitoring front and center.
This episode offers a practical look at what changes when AI leaves notebooks and enters production, governance, and the uneven realities of data work.
AI energy use is framed as an engineering problem with reliability, scalability, and sustainability at stake.
Data Science at Home treats the energy cost of large language models as an engineering tradeoff, not a vague concern. By drawing on blockchain comparison, proof of work, and quantization, it makes scale and efficiency feel concrete, while staying focused on the need to preserve…
Enterprise AI, with the hype pressure-tested.
Tech career stories, recapped warmly and turned into advice.
Francesco Gadaleta interviews Josh Miramant, CEO of Blue Orange Digital, about scaling data-driven businesses and the critical decisions founders face.
A forest after a storm sets the pattern.
Apple's privacy pitch has a control problem, as the host tells it.
This episode discusses how the use of AI tools like ChatGPT for productivity is leading to widespread, self-inflicted corporate espionage as employees feed confidential data into these systems.
Francesco discusses the concept of programmable money and how it could transform ownership and control of currency, using the analogy of a grandmother's purse to illustrate the loss of unconditional ownership.
Francesco explains the technical process behind serving large language models at scale, demystifying the magic of AI responses.
AI services get expensive fast.
Francesco Gadaleta brings Josh Miramont of Blue Orange Digital into a grounded chat about why AI and data systems buckle when notebook experiments, loose library versions, and production expectations get mixed together.
The episode opens on the energy cost of large language models.
Francesco interviews Mike Yeager, VP of Engineering at MetaRouter, about how enterprises can manage customer data privacy and compliance using first-party data routing.