Software Engineering Radio - the podcast for professional software developers

Birol Yildiz, CEO and co-founder of iLert, joins host Kanchan Shringi to explore how iLert built an AI SRE — an autonomous agent for handling production incidents — and what the experience revealed about building AI agents in the real world. Birol explains why incident response is a fundamentally agentic problem, where the unpredictability of novel incidents makes rule-based runbooks insufficient and reasoning models essential. He describes how the AI SRE evolved from an early browser-based approach to its current architecture, built around two key ingredients: reasoning models and the Model Context Protocol.
The conversation examines the four layers of the AI SRE in depth: an orchestration layer that routes requests and abstracts model providers; a knowledge layer built on plain text memory and agentic search rather than vector databases; an evaluation framework based on recorded live investigations replayed against new model versions; and a human-in-the-loop constraint layer. The episode concludes with practical advice for teams building agents: own your context completely, avoid off-the-shelf frameworks that obscure what enters the model, and get out of the way of the reasoning model rather than over-prescribing its steps.

Will Sentance, educator and co-founder of Codesmith, joins SE Radio's Adi Narayan to discuss the evolution of JavaScript and modern best practices. They begin with JavaScript's origins as a simple scripting language and its growth into the backbone of modern web development, highlighting the core theme of the "don't break the web" constraint. The requirement that JavaScript must remain backward-compatible has shaped everything from naming decisions (e.g., flat instead of flatten) to the introduction of Symbols as a collision-safe way to extend objects.
Will explains how the TC39 group uses the open-source community as a filtration system, absorbing user land patterns (like those from Lodash or Moment) into the standard library only once demand is proven. The upcoming Temporal API is highlighted as a major win for native date/time handling. On the engine side, Will discusses the shift toward monomorphic object shapes in the V8 JavaScript engine for better just-in-time (JIT) compiler performance, and how developers can now write more engine-aware code. The conversation also touches on LLMs in coding: Will's view is that AI tools are useful but risk atrophying developers' under-the-hood understanding, which remains essential for debugging complex, production-scale systems.

In this episode, host Amey Ambade sits with Eric Tschetter, co-founder of Apache Druid and Chief Architect at Imply, to dissect the critical move toward Decoupling Observability. To begin, they define three pillars—logs, metrics, and traces—and consider why the rise of microservices has made traditional, tightly coupled stacks a major source of pain. Such coupled systems can lead to issues such as vendor lock-in, prohibitive scaling costs, and operational complexity.
Drawing parallels to the Business Intelligence world's separation, Tschetter presents an architectural solution with four distinct layers: Ingest/Route, Data Storage, Query/Compute, and Visualization. This framework aims to provide flexibility to combat the limitations of monolithic observability tools. The conversation moves into the practical challenges and significant benefits of this decoupled model, focusing heavily on data portability and the role of technologies such as OpenTelemetry in standardizing schemas so that data can flow freely between multiple back-ends. A significant portion of the discussion is dedicated to the Query/Compute layer, specifically how Apache Druid addresses the unique demands of real-time analytics on observability data, including indexing strategies and unifying results across hot and cold storage. They also delve into operational survival, covering critical topics like smart sampling to preserve high-value signals, best practices for buffering and backpressure, and the governance models required for multiple teams to safely access the same data lake.
The episode concludes with an honest look at the complexity trade-offs and a roadmap for organizations considering a migration from a coupled vendor stack.

Martin Kleppmann, Associate Professor at the University of Cambridge and author of the best-selling O'Reilly book Designing Data-Intensive Applications, talks to host Adi Narayan about local-first collaboration software. They discuss what the term means, how it leads to simpler application architectures compared to the cloud-first model, and the benefits to developers and users from keeping all of their data on their own devices.  Martin goes into detail about how applications can synchronize data with and without a server, as well as conflict-resolution techniques, and the open-source library Automerge, which implements CRDTs and developers can use out-of-the-box. He also clarifies what kinds of applications would be suitable for the local-first approach. In the context of AI, they discuss vibe coding, local-first apps, and how the conflict-resolution work that enables data to be synchronized between users can also work with human-AI collaboration.

Sahaj Garg, co-founder and CTO of Wispr, a voice-to-text AI that turns speech into polished writing, talks with host Amey Ambade about designing systems for the ambiguity that's inherent in human input (text, voice, multimodal). Sahaj focuses on concrete architectural and training strategies for building robust AI systems. This episode examines the problem of ambiguity, where it shows up, building robust systems, personalization, communicating uncertainty, and evaluation. The conversation starts by exploring the difference between inherent and reducible ambiguity, major categories of ambiguity including lexical, syntactic, and pragmatic, and the additional sources of ambiguity in voice, such as homophones and accents. Garg details how to build systems through model training, including providing additional context and constructing datasets for good annotation. They discuss personalization with a focus on "revealed preferences"—learning from user behavior without explicit feedback—and fighting the problem of AI writing that "regresses to the mean." Finally, they consider how to communicate uncertainty to users without degrading the experience, as well as methods for evaluating ambiguity resolution through offline and online signals.