Tech on the Rocks

Join Kostas and Nitay as they speak with amazingly smart people who are building the next generation of technology, from hardware to cloud compute. Tech on the Rocks is for people who are curious about the foundations of the tech industry. Recorded primarily from our offices and homes, but one day we hope to record in a bar somewhere. Cheers!

Tech on the Rocks is a technology podcast hosted by Unknown Host, with 33 episodes on record and a Required Pod Score of 80.

Unknown Host hosts Tech on the Rocks, a technology show with 33 episodes published.

Our AI reads these to draft pitches. Use them as grounding for a pitch that cites a real guest and a specific topic.

• 

  Episode #33

  OCSF, Schemas, and the Convergence of Security and Data Engineering with Zach Schmerber

  Sep 11, 202658 min

  Zach joins Kostas and Nitay to trace how security data turned into a full-blown data engineering problem. He walks through his path from hospital networks with a sprawling attack surface, into tightly regulated banking, and then into big tech, where petabyte-scale volumes broke the Splunk-style stack he was hired to optimize and pushed him toward columnar storage and table formats like Iceberg and Delta. From there the conversation digs into OCSF, the Open Cybersecurity Schema Framework: why a producer-side schema would shrink the mountain of cleanup work defenders do today, how an unusually open community formed around it in a famously secretive industry, and why security schemas have to standardize semantics and values, not just structure. Zach, Kostas and Nitay also get into the overlap between security and observability stacks, the unknown-unknowns problem that forces teams to store data they may never query, the case for a metrics-plus-security semantic layer, and what changes when agents both need clean context and become the thing you have to monitor. Chapters 00:00 Introduction and Zach's background in cybersecurity 01:09 Early days in healthcare cybersecurity and data detection 02:19 The rise of Splunk and schema on read in security 03:07 Moving from healthcare to banking and regulated environments 04:21 Transition to big tech and handling massive data volumes 08:35 How Splunk's distributed compute works and scaling challenges 10:45 The shift towards modern data stacks and table formats 11:32 Introduction to the Open Cyber Security Schema Framework (OCSF) 14:00 Community and collaboration in cybersecurity data standards 15:36 The future of data schemas and standardization in security 20:59 Behavioral analytics and threat modeling in cybersecurity 23:28 The role of schemas in reducing security data complexity 28:01 Convergence of security data and observability 34:46 The impact of LLMs and AI on cybersecurity practices 50:46 The future of OCSF and standardization in cybersecurity

  ListenShow Notes
• 

  Episode #32

  Feeding the Agents: Fast Structured Data Retrieval with Arrow — Ian Cook, Co-founder & CEO of Columnar

  Aug 4, 20261h 7m

  In this episode, we talk with Ian Cook, co-founder and CEO of Columnar and a member of the Apache Arrow Project Management Committee, about ADBC (Arrow Database Connectivity) and why the way applications connect to databases is overdue for a rethink. Ian traces the history from the row-oriented database APIs of the 1990s, ODBC and JDBC, to today's world where nearly every analytic database and destination tool is columnar under the hood. He explains why keeping data in a columnar format end to end, using Apache Arrow, can deliver 10x to 100x speedups by turning a CPU-bound conversion problem back into a fast, network-friendly one, and why ADBC's flexible driver model works even against row-oriented systems like Postgres. We then dig into the AI angle: how fast structured data retrieval matters more than ever when agents reason in milliseconds and bottleneck on tool calls, why today's models still lean on tool calls to understand tabular data, and what tabular foundation models might change. Along the way we cover Parquet, DuckDB, Snowflake, string view, the DeWitt clause and benchmarking culture, and Arrow's philosophy of growing an ecosystem by building consensus rather than enemies. Ian also shares how to try ADBC yourself with the dbc CLI, a UV-inspired installer that makes it easy to install drivers for 20+ databases (columnar.tech/dbc). Chapters 00:00 Introduction to Ian Cook and Columner 01:53 Understanding ADBC and Its Relationship with Arrow 10:10 The Need for Columnar Paradigms in Database Connectivity 14:37 Exploring Use Cases for ADBC in Modern Applications 18:41 Performance Impacts of ADBC in Various Systems 27:38 Integrating ADBC with AI and LLMs 37:26 The Future of ADBC and Its Role in Data Infrastructure

  ListenShow Notes
• 

  Episode #31

  Falling Into Databases: The DuckDB Story with Hannes Mühleisen

  Jul 7, 202654 min

  In this episode, we sit down with Hannes Mühleisen, co-founder and CEO of Duck Labs and co-creator of DuckDB, for a wide-ranging conversation about how a self-described outsider ended up reshaping the analytics database world. Hannes shares how he accidentally fell into databases after moving to Amsterdam, why that outsider perspective helped him spot the “warts” everyone else had accepted, and his theory that traditional databases were “sold on the golf course” rather than built for the people who actually use them. We dig into the origins of DuckDB: the decision to throw away a working prototype and rebuild from scratch, the untapped gap between Excel/pandas and Spark, and why in-process analytics unlocked a whole new class of users. Hannes also explains the “venture communism” reasoning behind open-sourcing a project built with Dutch tax dollars, how a single Hacker News post lit the fuse, and the philosophy of solving one real user problem at a time. Finally, we get into the latest work — DuckLake, the Quack client-server protocol, the “never give up, never surrender” approach to queries that refuse to crash, and what changes when the entity querying your database shifts from humans to machines. Chapters 00:00 Introduction to Hannes and His Journey into Databases 02:35 The Outsider Perspective in Database Development 05:12 The Evolution of Database Sales and User Experience 08:01 The Rise of Open Source in Database Systems 10:56 Building DuckDB: Challenges and Innovations 13:31 The Gap in OLAP Systems and the Emergence of DuckDB 16:05 Creating a New Market for Analytics with DuckDB 19:05 The Impact of Open Source and Academic Roots 21:57 The Philosophy Behind Open Sourcing DuckDB 28:02 The Value of Public Funding in Research 29:46 Open Source Strategy and Market Credibility 31:39 Launching DuckDB: Initial Reactions and Strategies 35:07 Learning from User Feedback and Market Needs 38:38 The Challenges of Database Development 42:34 Innovations in Client-Server Protocols 47:02 The Evolution of Client-Server Protocols 52:57 Future Aspirations for DuckDB and Community Engagement

  ListenShow Notes
• 

  Episode #30

  Rebuilding the Robot Stack: Why Robotics Needs a New Real-Time OS with Guillaume Binet (Copper Robotics)

  Jun 19, 202656 min

  In this episode, Nitay and Kostas sit down with Guillaume Binet, founder of Copper Robotics, to dig into one of the most overlooked problems in modern robotics: the software stack that runs the robots. Guillaume traces his journey from tinkering with old computers in the 80s and 90s, to telecommunications, to the dot-com era, and finally into more than a decade in robotics, including roles at Trilio, Google, Motional, Argo AI, and as CTO at Skyways building long-range carbon-fiber logistics drones. Along the way, he discovered a recurring gap: while cloud software has matured into a rich ecosystem of tools and frameworks, the software powering real-world robots is still surprisingly broken. That insight led him to start Copper Robotics, where he is building a new operating system and runtime designed specifically for real-time robots. We talk about: Why robots are fundamentally different from cloud services and laptops, and what "real-time" actually means when sensors, perception, and actuation all have to meet a fixed time budget The mismatch between microservice-style architectures (and ROS) and the constraints of an autonomous system that has to react in milliseconds How Copper builds a statically described, deterministic operating system around your robot, with a scheduler tailored to the exact shape of the system Running the same Rust codebase across heterogeneous compute, from Linux hosts to bare-metal MCUs, with a single self-contained executable Why TCP/IP is usually the wrong answer inside a robot, and how to think about latency, bandwidth, and dropped data in a real-time context The trade-off between compile-time and runtime flexibility, and where Copper draws the line versus ROS Safety certification for autonomous systems (ISO 26262, aerospace, medical) and why 100% deterministic replay is a game-changer for proving that what you tested in simulation is what runs on the robot The community forming around Copper: students, new robotics startups, and teams who have already "hit the wall" with existing tooling A fun detour into restoring 80s and 90s computers, floppy disks, and the lost art of magnetic media If you have ever wondered why we see so many incredible robot demo videos but so few robots actually deployed in the real world, this conversation is for you. Learn more about Copper Robotics and join the community via their open source project and Discord.

  ListenShow Notes
• 

  Episode #29

  Physical AI and the Future of Robotics with Sergey Arkhangelskiy of Positronic

  Jun 5, 202652 min

  In this episode, Nitay and Kostas sit down with Sergey Arkhangelskiy, founder of Positronic, to dig into the state of physical AI and what it will really take to bring general-purpose robotics into the real world. Sergey shares his journey from a decade at Google Search, where he worked on ranking and helped build the "Tetris" layer that unified web, image, and map results, to co-founding Wanna (a computer vision AR company acquired by Farfetch), and most recently launching Positronik to focus on robotics and physical AI. The conversation explores why robotics is approaching but has not yet hit its "GPT-3 moment," the data and hardware bottlenecks that make physical AI fundamentally harder than LLMs, and why measurement and evaluation matter so much. Sergey walks through Positronik's newly released Physical AI Leaderboard (fail.ai), which benchmarks open-source vision-language-action (VLA) models on real hardware using production-grade metrics like throughput (units per hour) and mean time between failures, rather than simple success rates. They also discuss why commercial and industrial applications (manufacturing, logistics, pick-and-place) are likely to lead before household robots, the economics of automation and the "cost of intelligence," the role of human-in-the-loop systems, latency and cloud-vs-edge tradeoffs for running VLA models, and the growing importance of open source in both robotic software (ROS) and hardware (OpenArm). Topics covered: Sergey's path from Google Search ranking to robotics Why physical AI is harder than LLMs: data, hardware, and the real world The Physical AI Leaderboard and how it evaluates VLA models on real robots Throughput and mean time between failures as production metrics Why commercial use cases will lead household robotics The economics of automation and the "cost of intelligence" Human-in-the-loop and the realistic path to full automation Cloud vs. local inference, latency, and bandwidth constraints on the factory floor The role of open source in robotics hardware and software What the ecosystem needs next to accelerate adoption

  ListenShow Notes

Recent guests on Tech on the Rocks. Study who booked and why before you pitch.

Based on semantic analysis of episode topics and host coverage, this show is a strong guest fit for executives in:

Industry fit is computed by PitchCentric using vector embeddings of the show's episode catalog.