The CTO Advisor

5 episodes scanned. 100% of ad slots promote the show's own products.

Recent Episodes

1. Episode 704: When You Can’t Trust Your Software Updates

   Nov 3, 2026

   Co-host Keith Townsend joins Calvin Hendryx-Parker to discuss a week of major security disclosures, including remote shell exploits in PHP, a certificate authority with crackable keys, a Brotli RCE affecting Python, and a second CI pipeline compromise at LiteLLM. The conversation covers whether these are developer or operator problems and how to apply zero trust when vendor security updates can't be fully trusted. They discuss package cooldown periods in uv and pip, least privilege for build pipeline keys, and Keith's experience with a coding agent that closed every app on his Mac, leading to advice on scoped, time-bound agent permissions for teams without dedicated security staff.

2. Episode 703: Your AI Capacity Plan Depends on More Than GPUs

   Oct 26, 2026

   Calvin Hendryx-Parker, co-founder and CTO of Six Feet Up, joins Keith Townsend to discuss whether the AI bubble is bursting, focusing on Oracle's Stargate data center commitments and funding concerns. The conversation centers on power as the key constraint in AI infrastructure, including colocation shifts from floor space to power, GPU supplier continuity, and cloud GPU access. They close with the risk of sticky managed services like CoreWeave's Weights & Biases and the case for planning hybrid workflows, not just hybrid cloud.

3. Episode 702: Who Gave Your AI Agent That Authority?

   Oct 20, 2026

   Calvin Hendryx-Parker, co-founder and CTO of Six Feet Up, joins host Keith Townsend to examine who grants authority to AI agents, using the reported AWS AgentCore credential case and a Codex CLI update that leaves a self-updating background daemon running after exit as starting points. The conversation covers seeded authority, enterprise persistent memory, Slack bots in federated channels, and accountability when an agent pursues a goal no one instructed. Calvin recommends granular, explicit roles on agent containers using the Paperclip meta harness with a Kubernetes agent operator, and Keith proposes four governance tests: authorization boundaries, state exportability, behavior traceability, and memory independence.

4. Episode 701: Everpure’s Rob Lee on GPUs, Data Gravity, and KV Cache

   Oct 13, 2026

   Rob Lee, Chief Technology and Growth Officer at Everpure (formerly Pure Storage), discusses how CTOs can tell a true compute shortage from a data problem, arguing that task latency and accuracy should drive infrastructure decisions more than GPU utilization. The conversation covers data gravity, why AI agents working in transactional flows pull compute toward enterprise data, and the need to reason across silos such as SAP, CRM, and unstructured data. Lee also explains KV cache offload as part of a broader memory hierarchy for inference and outlines Everpure's current AI infrastructure and AI data readiness offerings.

5. Episode 700: A New Co-Host, TypeSafe’s Jev, and Qwen3.8 at Home

   Oct 1, 2026

   Episode 700 marks a reboot of the CTO Advisor podcast with new co-host Calvin Hendryx-Parker, CTO and co-founder of Six Feet Up, joining Keith Townsend. The episode explores two major AI topics: Jev, TypeSafe AI's closed-weight classifier that delivers typed answers in milliseconds at low cost, and its implications for data privacy; and Qwen3.8, an open-weight model from Alibaba that runs efficiently locally, including open-source classifier alternatives like JevKED5. The hosts discuss the tradeoffs between classifier models, fine-tuning, RAG pipelines, and the practical considerations of local versus cloud inference for CTOs.

6. Layer2C Labs 001: Borrow the Vendor’s Plumbing, Not Its Judgment

   Aug 24, 2026

   Keith Townsend presents Layer2C Labs 001, a technical deep dive into building a retrieval-augmented generation (RAG) pipeline over 13 years of Tech Field Day transcripts (237,000 segments). He compares a hybrid architecture—AWS managed data plane (Titan embeddings, S3 Vectors) with local NVIDIA DGX Spark compute—against fully cloud-based inference, revealing that while managed services are cheaper and faster, they hide critical control points, particularly in data chunking and segmentation strategy.

7. Deterministic AI Isn’t the Goal. Deterministic Systems Are.

   Jul 29, 2026

   Adam Jacobs, CEO of Swamp Club, discusses the misconception around deterministic AI and why the real goal is building deterministic systems. Drawing from his experience founding Chef during the DevOps movement, Jacobs explores how AI agents can be architected to produce reliable, repeatable outcomes by separating deterministic workflows from probabilistic reasoning, using testing and validation to ensure consistency.

8. Your GPUs Aren't Slow. Your Data Path Is

   Jul 10, 2026

9. Where HPE Adds Value to NVIDIA CPUs

   Jul 9, 2026

10. Not All NVIDIA AI Factories Are the Same

    Jul 8, 2026