Practical MLOps for Network Operations at Uber

Host Philip Gervasi talks with Uber's Vishnu Acharya about how Uber applies machine learning and MLOps to network operations at hyperscale. Vishnu explains Uber’s intentionally simple network design across on-prem and multi-cloud, then shares practical machine learning use cases like predictive capacity planning, hardware failure rate-tracking, and alert correlation to reduce noise and speed mitigation. They also discuss organizational issues, including building blended network/software teams, partnering with internal ML groups, and focusing on service-level outcomes over hype.

A new wave of civil unrest in Iran has triggered one of the most severe communications crackdowns the country has seen, with disruptions extending beyond the internet to voice, SMS, and Starlink. Host Philip Gervasi is joined by Amir Rashidi (Miaan Group / Filterwatch) and Kentik’s Doug Madory to break down how the shutdown unfolded, what the data shows as connectivity collapses, and why attribution is tricky in the fog of a rapidly evolving crisis. They discuss Iran’s National Information Network (the domestic “intranet”), how shutdown tactics have grown more targeted and sophisticated over time, and what makes this event different.

Host Philip Gervasi is joined by Andrés Azpúrua (Executive Director, VE sin Filtro / “Free and Secure Online”) and Kentik’s Doug Madory to examine what internet visibility can, and can’t, tell us during a fast-moving political crisis in Venezuela. They discuss Venezuela’s “baseline” reality of brittle infrastructure and recurring power outages, alongside long-standing, regulator-mandated censorship targeting independent media, social platforms, and even exchange-rate information. Using multiple data perspectives, the episode explores how traffic patterns, localized outages, and BGP noise can be misread without ground truth, and why careful attribution matters.

Host Phil Gervasi talks with Chris Grundemann, cofounder of the Network Automation Forum and AutoCon, about why network automation still isn’t fully adopted after more than a decade of hype. They dig into the real blockers behind the tech: skills gaps, engineer identity and culture, organizational structure, and the broader “sociotechnical” system around NetOps. Chris also shares how the AutoCon community has evolved over five events, the growing role of observability and AI in automation, and what’s next for NAF and AutoCon in Europe.

Kentik’s Mav Turner joins host Phil Gervasi to go beyond chatbot hype and dig into real AI reasoning for network operations. They discuss how Kentik AI Advisor uses network intelligence, hybrid RAG, and tool-calling to troubleshoot issues, optimize cost, and democratize access to network expertise. Along the way, they cover architecture, data governance, model evaluation, and why AI has to be built into an observability platform itself, not bolted on.

Host Philip Gervasi talks with Kentik pre-sales engineer Sean McGinley about what it really means to work in pre-sales. They unpack the various titles associated with those roles (e.g., solutions engineer, solutions architect) and discuss how pre-sales bridges the gap between technology and business. Learn how these roles balance hands-on engineering with customer relationships. Along the way, they share personal stories, lessons learned from the field, and why pre-sales can be one of the most rewarding career paths in tech.

Host Philip Gervasi and Kentik Field CTO Justin Ryburn preview the upcoming AutoCon 4 conference, one of the premier events for the networking industry. They discuss the growth of the Network Automation Forum, the evolution of the AutoCon community, and the event’s focus on real-world network automation, observability, and AIOps. Justin previews Kentik’s hands-on AutoCon 4 session, “Building Smarter Observability with Network Intelligence.”

Kentik CPO Mav Turner joins host Philip Gervasi to cut through the AI hype in NetOps. They discuss where ML and LLMs actually help—anomaly detection, root cause analysis, and agent-driven runbooks—and where deterministic methods still win. Join us for real talk on data pipelines, telemetry quality, model evaluation, human-in-the-loop guardrails, and the build-vs-buy trade-offs that transform network noise into informed decisions.

Doug Madory joins us to unpack the recent Red Sea submarine cable cuts and how Kentik’s Cloud Latency Map revealed the global impact in real-time, offering critical insight into cloud performance, interconnectivity, and internet resilience.

Lauren Basile joins us to show how traffic-aware cost intelligence turns spreadsheet guesswork into one-click, per-slice cost estimates across customers, ASNs, and CDNs. Learn about the SNMP plus contracts foundation, the flow-data leap, and how NetOps teams use cost-per-Mbps and path insights to optimize spend, pricing, and margins.