Day Two DevOps
Packet Pushers
Join hosts Ned Bellavance and Kyler Middleton as they dive deep into the challenges of DevOps from the perspective of seasoned practitioners. You'll hear from expert guests—technical leaders, trainers, and consultants with years of hands-on experience—discussing the nuances of DevOps. From AWS to Azure, networking to security, automation to modern cloud environments, each weekly episode equips you with the insights to confidently address tech and business challenges such as resilience, cost management, and performance. Whether you want to hone your skills today or prepare for what’s coming next, Day Two DevOps cuts through the vendor fog to guide you through a shifting IT landscape.
Episodes
Mentioned books
Nov 4, 2026 • 43min
D2DO318: Navigating Open-Source Model Providers
Kyler and Ned are joined by Mo Moustafa, creator of Ollybot, to explore the complexities of running open-source models through OpenRouter. Together they delve into model tuning, experiments with model providers, the roiling chaos that is the current state of inference as a service, and more. Packet Pushers by Meter Episode Links: Watch this episode... Read more »
5 snips
Oct 28, 2026 • 49min
D2DO317: Building the Operational Reality for Sandboxed Agents in Production (Sponsored)
Michael Levan, a cloud-native and AI infrastructure practitioner, and Nathan Pearce, a specialist in agentic systems and secure operations, explore what it takes to run AI agents in production. They discuss stateful runtimes, defense-in-depth sandboxing, blast radius, observability, scale-to-zero, snapshots, and enterprise use cases. They also introduce Agent Substrate and Kagent for governed, efficient Kubernetes-based agent operations.
Oct 21, 2026 • 45min
D2DO316: Local AI Foundations
A practical tour of small language models, open-weight customization, and the trade-offs between local and frontier AI. Explore running models with LM Studio and Ollama, hardware needs, quantization, training versus inference, and the security challenges of AI agents and tool access.
Oct 7, 2026 • 47min
D2DO315: Running Agents Securely at Scale
Michael Levan, AI architect at Solo.io and cloud-native educator, joins Ned and Kyler to explore running AI agents securely at scale. They unpack specialized gateways, semantic model routing, agent identity, sandboxing, and observability. The conversation also covers Kubernetes as an agent runtime, local models, cost control, and measuring agent efficiency without misleading metrics.
Sep 23, 2026 • 46min
D2DO314: Backups SaaS – Safety as a Service
Jim Jones, a data protection architect at 11:11 Systems, joins Ned and Kyler to explore backup strategy in the hybrid-cloud and AI era. They discuss ransomware, cloud automation’s destructive power, AI agents exposing sensitive data, permission scoping, DSPM, zero trust, and immutable backups. They also examine pipeline safeguards and Jim’s Obsidian-based AI knowledge workflow.
Sep 16, 2026 • 46min
D2DO313: Operations is a Reconciliation Loop (Sponsored)
Paul Delory, Gartner research vice president and former virtualization engineer, explores operations as a reconciliation loop. Topics include continuous operations, GitOps, immutable infrastructure, and managing everything from servers to network devices. He also examines how AI agents need deterministic guardrails, policy as code, and safer Terraform and Ansible workflows, plus the rise of platform engineering and Kubernetes.
Sep 9, 2026 • 43min
D2DO312: Networking at Scale: AWS Transit Gateway War Stories
Devender Singh, a Principal Cloud Architect and AWS networking specialist, shares hard-won lessons from managing 95,000 hosts. He explores M&A migration chaos, stakeholder planning, overlapping networks, rollback strategies, and consolidating duplicate infrastructure. The conversation also compares AWS Transit Gateway with Cloud WAN, regional firewall architectures, financial-sector requirements, and the surprising costs of firewall licensing.
27 snips
Aug 26, 2026 • 50min
D2DO311: Code Review Is Dead – Long Live Peer Review
Dylan Ratcliffe, founder and CEO of Overmind Technology and DevOps entrepreneur who builds AI-driven infrastructure simulation tools. He discusses how AI changed review workflows and why code review norms need rethinking. Short takes on treating AI as an auditable machine, shifting reviews to human-authored implementation plans, and cultural rules like producing working code before asking for signoff.
19 snips
Aug 12, 2026 • 42min
D2DO310: Developing Efficient AI Workflows
Tyler Lynch, Outbound CTO at IBM working on automation, explains shifting from tokenmaxxing to valuemaxxing. He covers efficient AI agent workflows, using persistent memory to avoid context bloat, secure sandboxing and least-privilege access, and routing tasks across models with parallel verification for reliability.
Jul 29, 2026 • 58min
D2DO309: Packet Pushers Assemble! Bridging the Telemetry Divide
William Collins, network-to-cloud engineer and podcast host, and Scott Robon, seasoned network ops practitioner and podcaster, tackle telemetry and observability across networking and cloud. They talk about SSH scraping and legacy network telemetry. They cover cloud logging scale, OpenTelemetry adoption, vendor fragmentation, streaming telemetry, and how AI might help normalize and analyze noisy data.
