# Code generation Published articles for Code generation. This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles. ## DACA-GRPO: Denoising-Aware Credit Assignment for Reinforcement Learning in Diffusion Language Models DevFeed: [DACA-GRPO: Denoising-Aware Credit Assignment for Reinforcement Learning in Diffusion Language Models]() Original publisher: [Read original article]() Published: 2026-09-16T00:00:00Z Content type: article Language: en Sources: [Apple Machine Learning Research]() Topics: [grpo](), [Reinforcement learning](), [large-language-models](), [LLM Techniques](), [Code]() Tags: [code-generation](), [denoising](), [diffusion](), [grpo](), [json-schema](), [optimization](), [reinforcement-learning]() ### AI overview The article introduces DACA-GRPO, a denoising-aware enhancement for GRPO-style reinforcement-learning trainers used with diffusion language models. It uses per-token denoising progress scores and stratified masking likelihood to improve credit assignment and reduce bias in likelihood estimates. Across seven benchmarks, the method reports gains in mathematical reasoning, code generation, constraint satisfaction, and JSON Schema adherence. ### Source excerpt Diffusion large language models are a compelling alternative to autoregressive models, yet existing RL methods for diffusion treat all denoising steps as equally important and rely on biased, high-variance likelihood estimates. We identify two fundamental weaknesses: the absence of temporal credit assignment across the denoising trajectory, and the systematic bias of mean-field likelihood estimates used for policy optimization. To address these, we propose Denoising-Aware Credit Assignment for GRPO (DACA-GRPO), a lightweight, plug-and-play enhancement for any GRPO-style trainer. DACA-GRPO... ## Article: When Spec-Driven Development Pays Off DevFeed: [Article: When Spec-Driven Development Pays Off]() Original publisher: [Read original article]() Author: Nitin Garg Published: 2026-09-10T09:00:00Z Content type: article Language: en Sources: [InfoQ]() Topics: [AI-assisted coding](), [AI Development](), [Software Engineering](), [code productivity]() Tags: [ai](), [ai-assisted-coding](), [ai-coding](), [ai-development](), [ai-ml-data-engineering](), [architecture](), [article](), [artificial-intelligence](), [code-generation](), [generative-ai](), [ml-data-engineering](), [productivity](), [reasoning](), [review](), [security](), [software-development](), [spec-driven-development](), [when-spec-driven-development-pays-off]() ### AI overview The article argues that AI-assisted coding shifts the main constraint from writing code to verifying it. It presents specification-first development as a governance approach for hard, multi-constraint work, while noting its time and cost and warning that apparent gains may instead come from reasoning. ### Source excerpt AI coding assistants have become a core part of software development. AI-generated code has shown productivity gains, but it's also contributing to security weaknesses and familiar bug patterns. In this article, author Nitin Garg highlights the bottleneck has moved from code generation to code verification, and how to detect & mitigate it when the AI-generated behavior diverges from the intent. By Nitin Garg ## Introducing Agent-Ready Code Repository & AI Code Review DevFeed: [Introducing Agent-Ready Code Repository & AI Code Review]() Original publisher: [Read original article]() Author: Juveria Kanodia Colin Chartier Published: 2026-08-27T00:00:00Z Content type: release Language: en Sources: [Harness Blog]() Topics: [Artificial Intelligence](), [AI-assisted coding](), [Pull Request](), [Code review]() Tags: [ai](), [ai-agents](), [ai-code-review](), [code-generation](), [code-review](), [software-delivery]() ### AI overview Harness introduces an agent-ready Code Repository and built-in AI Code Review to manage AI-generated code at high volume. The capabilities use risk-based diffs and scoped agent permissions to support governed software delivery before code merges. ### Source excerpt Legacy SCMs can't handle agent-scale code volume. Learn how Harness Code Repository and built-in AI Code Review handle AI-generated code at scale using risk-bas | Blog ## MSVC Build Tools Preview updates - August 2026 DevFeed: [MSVC Build Tools Preview updates - August 2026]() Original publisher: [Read original article]() Author: Eric Brumer Published: 2026-08-13T15:01:10Z Content type: article Language: en Sources: [C++ Team Blog]() Topics: [MSVC](), [C++](), [Compiler](), [toolchain](), [Code generation](), [Optimization](), [Visual Studio]() Tags: [c-plus-plus](), [code-generation](), [diagnostics](), [features](), [ide](), [msvc](), [optimization](), [release](), [tools](), [updates](), [visual-studio](), [visual-studio-2026]() ### AI overview The article summarizes August 2026 updates to the MSVC Build Tools Preview, targeting the v14.52 release. It covers compiler conformance, diagnostics, C++ modules, code generation, optimization, and ARM64 improvements. ### Source excerpt Try the latest updates to MSVC Build Tools Preview The post MSVC Build Tools Preview updates - August 2026 appeared first on C++ Team Blog. ## Building Internal Flashboards in Slack with AI DevFeed: [Building Internal Flashboards in Slack with AI]() Original publisher: [Read original article]() Author: WorkOS Published: 2026-08-10T16:15:41Z Content type: article Language: en Sources: [WorkOS Blog]() Topics: [Slack](), [dashboards](), [Artificial Intelligence](), [Claude](), [data](), [API](), [HTML]() Tags: [agent](), [ai](), [api](), [claude](), [code-generation](), [html](), [leaderboards](), [reporting](), [slack]() ### AI overview WorkOS describes Flashboards, an internal reporting tool built as documents with live data connections and edited through AI agents in Slack. The article explains its HTML-based architecture, read-only data access, employee authentication, natural-language creation workflow, interactive filters, and versioned iteration. ### Source excerpt Flashboards is WorkOS's internal reporting tool: a doc with a live data connection, and agents as the only editing interface. Zero to 55 weekly readers in seven weeks. ## Agentic AI Development Tools: How to Build Your Stack DevFeed: [Agentic AI Development Tools: How to Build Your Stack]() Original publisher: [Read original article]() Author: Aaron Taylor Published: 2026-08-10T11:34:27Z Content type: article Language: en Sources: [Developer Experience & Platform Engineering Blog | Port]() Topics: [AI Development](), [developer tooling](), [AI-assisted coding](), [sdlc](), [CI/CD](), [observability]() Tags: [agentic-ai](), [agents](), [ai-agents](), [ai-development](), [ci-cd](), [code-generation](), [development-tools](), [observability](), [sdlc](), [software-delivery]() ### AI overview This guide explains how to choose and connect agentic AI development tools across the software development lifecycle. It distinguishes the agent layer, the agentic infrastructure layer, and the existing foundation of source control, CI/CD, and observability tools, then introduces criteria for evaluating tools and begins with code generation agents. ### Source excerpt Explore agentic AI development tools and learn how to build a practical stack for faster, more reliable AI software delivery. ## Keleusma Research Spike: Blocking Frequency as the Ordering Principle for Instruction-Set Coverage DevFeed: [Keleusma Research Spike: Blocking Frequency as the Ordering Principle for Instruction-Set Coverage]() Original publisher: [Read original article]() Author: Brendan Sechter Published: 2026-08-06T09:00:00Z Content type: article Language: en Sources: [Brendan A R Sechter's Development Blog]() Topics: [Compiler](), [Code generation](), [Code]() Tags: [article](), [code-generation](), [compiler](), [compilers](), [engineering](), [measurement](), [ordering](), [research](), [verification]() ### AI overview This case study examines how blocking frequency can guide the order of instruction implementation in Keleusma's compiler backend. It contrasts instruction-level coverage with whole-program compilability and describes a small measurement tool that exposed a flawed implementation plan. ### Source excerpt A compiler was 87 percent finished. It could not compile two thirds of the programs it was for. Both numbers are correct. The first counts individual instructions the compiler knew how to translate. The second counts whole programs that would actually go through. The gap between them is what this article is about, and the reason it exists is simple enough to state in one sentence. A program needs every instruction it uses, not most of them. One missing instruction out of a hundred stops the whole thing, exactly as one missing link stops a chain. That gap then destroyed a carefully reasoned plan. One working session before the measurement was taken, the author of this article had formally recommended what the next piece of work should be. The reasoning had no invalid step in it. The measurement showed the recommendation to be worth nothing at all, because the thing it would have unblocked does not occur even once in any program the compiler is meant to serve. The instrument that established this took about twenty minutes to build and two seconds to run. The article reports that, and then reports four errors made while writing it, all four of which ran in the direction of a more striking result, and one of which was committed inside the paragraph warning against the other three. What this is a case study of The setting is compiler engineering, and a reader who has never written a compiler can follow the argument, because the shape of the problem is not specific to compilers. The concrete project is Keleusma, whose compiler until now has emitted bytecode for a virtual machine, as described in the self-hosting strategy and its getting-started article. Native code generation is the step after that one, and it is where the ordering question first became expensive enough to measure. The lineage of the design sits in the stream-based compilers series and in the self-hosted silicon compiler. None of that background is needed to follow what follows, and the measurement stands ## Muse Spark 1.2 is now available on Vercel AI Gateway DevFeed: [Muse Spark 1.2 is now available on Vercel AI Gateway]() Original publisher: [Read original article]() Author: Jerilyn Zheng Published: 2026-08-05T00:00:00Z Content type: release Language: en Sources: [Vercel News]() Topics: [AI-assisted coding](), [vercel ai sdk](), [debugging]() Tags: [agent](), [ai-gateway](), [api](), [code-generation](), [debugging](), [muse](), [spark](), [vercel]() ### AI overview Muse Spark 1.2 is available on Vercel AI Gateway as a coding-focused model update, with stated improvements in code generation, debugging, codebase understanding, and iterative developer workflows. ### Source excerpt Muse Spark 1.2 from Meta is now available on AI Gateway. It is a coding-focused update to the previous Muse Spark model. While keeping its general capabilities, 1.2 ships with improvements in code generation, complex debugging, codebase understanding, and end-to-end developer workflows. The model is built for long-horizon work like generating whole repositories, building out large projects end to end, and sustaining iterative loops where it writes, compiles, profiles, and improves code over many rounds. To use Muse Spark 1.2, set model to meta/muse-spark-1.2 in the AI SDK: To use Muse Spark in a coding agent, run vercel ai-gateway coding-agents setup to connect Claude Code, Codex, OpenCode, or Pi, then select meta/muse-spark-1.2 inside the agent. AI Gateway provides a unified API for calling models, tracking usage and cost, and configuring retries, failover, and performance optimizations for higher-than-provider uptime. It includes built-in custom reporting, Zero Data Retention support, budgets for API keys, routing rules, and more. AI Gateway reflects provider pricing with no markup and does not charge a platform fee on inference, including on Bring Your Own Key (BYOK) requests. Try Muse Spark 1.2 in the model playground. Read more ## KotlinLLM is Going Open Source DevFeed: [KotlinLLM is Going Open Source]() Original publisher: [Read original article]() Author: Anastasia Birillo Published: 2026-07-28T07:50:41Z Content type: article Language: en Sources: [Kotlin : A concise multiplatform language developed by JetBrains | The JetBrains Blog]() Topics: [kotlinllm](), [Kotlin](), [Open Source](), [IntelliJ IDEA](), [Large Language Model](), [Code generation](), [AI-assisted coding](), [Software Engineering]() Tags: [code-completion](), [code-generation](), [engineering](), [intellij](), [intellij-idea](), [kotlin](), [kotlinllm](), [llm](), [llms](), [open-source](), [plugin](), [research](), [research-prototype](), [software](), [software-engineering]() ### AI overview KotlinLLM is an open-source research prototype and IntelliJ IDEA plugin for Kotlin/JVM projects. It introduces Smart macros whose generated Kotlin source evolves from runtime scenarios, allowing applications to delegate selected logic to an LLM without calling the model on every request or relying on a separate runtime agent. ### Source excerpt TL;DR KotlinLLM is now public. It's a research prototype for delegating runtime logic to an LLM from Kotlin code. Instead of calling an LLM on every request or running a separate agent, you can write an explicit Kotlin call. Its body is generated Kotlin source code, and that code is updated as your application hits [...] ## NVIDIA Nemotron 3 Ultra Leads Open Models on Accuracy and Efficiency in Agentic RTL Coding DevFeed: [NVIDIA Nemotron 3 Ultra Leads Open Models on Accuracy and Efficiency in Agentic RTL Coding]() Original publisher: [Read original article]() Author: Nirmal Kumar Juluru Published: 2026-07-27T00:45:00Z Content type: article Language: en Sources: [NVIDIA Developer](), [NVIDIA Technical Blog]() Topics: [coding](), [Verilog](), [AI-assisted coding](), [Nvidia](), [Artificial Intelligence](), [debugging](), [Benchmark](), [Code generation](), [Hardware](), [long-context]() Tags: [agentic](), [agentic-ai-generative-ai](), [ai](), [ai-agent](), [ai-agents](), [automation](), [benchmark](), [chip-design](), [code](), [code-generation](), [coding](), [debugging](), [developer-tools-techniques](), [efficiency](), [engineering](), [featured](), [hardware-semiconductor](), [llms](), [long-context](), [models](), [nemotron](), [nvidia](), [open](), [reasoning](), [top-stories](), [validation]() ### AI overview NVIDIA's article presents ACE-RTL and Nemotron 3 Ultra as a combined approach to agentic RTL coding. The workflow generates Verilog, runs simulations and other EDA checks, analyzes failures, and iteratively refines designs while maintaining debugging context. The CVDP benchmark evaluates accuracy and efficiency on realistic RTL generation, modification, debugging, and verification tasks. ### Source excerpt Modern chip design is increasingly limited by engineering time. Register transfer level (RTL) development and verification require specialized hardware... ## How Figma stays ahead of vulnerabilities with agents DevFeed: [How Figma stays ahead of vulnerabilities with agents]() Original publisher: [Read original article]() Author: Rohan Sharma; Liam Buchan; Dave Martin Published: 2026-07-23T21:23:00Z Content type: article Language: en Sources: [Figma Blog]() Topics: [Figma](), [Vulnerabilities](), [Pull Request](), [AI-assisted coding](), [Claude Code](), [codex](), [Security](), [Code]() Tags: [agentic](), [claude](), [claude-code](), [code-generation](), [codex](), [coding](), [cost](), [engineering](), [figma](), [pull-request](), [review](), [security](), [vulnerabilities]() ### AI overview Figma describes how it uses coding agents to prevent, detect, and fix vulnerabilities during code generation, pull request review, and historical code auditing. The article focuses on improving precision, recall, and trust, including a review system that combines Claude Code with Opus 4.8 and Codex with GPT-5.6 Sol. ### Source excerpt For the past year, agents at Figma have guarded code as it's written, reviewed every pull request, and audited a decade-old monorepo, all on one policy. ## 10 Platform engineering predictions for 2026 DevFeed: [10 Platform engineering predictions for 2026]() Original publisher: [Read original article]() Author: Luca Galante Published: 2026-07-23T05:40:01Z Content type: article Language: en Sources: [Platform Engineering Blog]() Topics: [Platform Engineering](), [Artificial Intelligence](), [Security](), [Architecture & Design](), [Vibe coding](), [Kubernetes](), [Terraform]() Tags: [2026](), [agentic](), [agentic-ai](), [ai](), [ai-agents](), [architecture](), [code](), [code-generation](), [coding](), [finops](), [iam](), [infrastructure](), [kubernetes](), [models](), [platform](), [platform-engineering](), [predictions](), [security](), [terraform](), [vulnerabilities]() ### AI overview The article presents ten predictions for platform engineering in 2026, focusing on the convergence of AI, security-by-design, developer experience, observability, and FinOps. It argues that platform teams will increasingly govern autonomous AI agents, secure AI-generated infrastructure code, and use AI-driven optimization to re-architect systems for cost and latency targets. ### Source excerpt Explore the 10 platform engineering predictions for 2026. See how AI, FinOps, and new security models will fundamentally redefine the platform team's role and architecture. ## Announcing Capital One's 2026 UIUC AI Awardees DevFeed: [Announcing Capital One's 2026 UIUC AI Awardees]() Original publisher: [Read original article]() Author: Capital One Tech Published: 2026-07-21T14:45:44Z Content type: release Language: en Sources: [Capital One Tech]() Topics: [Artificial Intelligence](), [ai safety](), [Generative AI](), [Large Language Model](), [Code generation](), [Cybersecurity]() Tags: [2026](), [academic-research](), [agentic-ai](), [agents](), [ai](), [ai-research](), [ai-safety](), [code-generation](), [cybersecurity](), [generative-ai](), [large-language-models-llms](), [research](), [science]() ### AI overview Capital One announces its 2026-2027 research and fellowship awardees from the University of Illinois. The featured projects address LLM reasoning faithfulness, code reasoning, imaginative LLM agents, and agentic AI safety. ### Source excerpt Meet the University of Illinois researchers and fellows advancing Agentic AI through our academic partnership.Announcing the Center for Generative AI Safety, Knowledge Systems, and Cybersecurity (ASKS) 2026-2027 Capital One Research Awardees from the University of Illinois As we look toward the 2026-2027 academic year, we continue to partner with institutions that lead the global conversation on the future of intelligence. We are thrilled to announce this year's cohort of research and fellowship awardees from the University of Illinois, whose pioneering work addresses the most critical frontier in technology today: Agentic AI. From the way machines "think" and "imagine" to the hardware that powers them and the safety protocols that govern them, these five projects represent Capital One's holistic push toward AI that is not only powerful but also faithful, safe and creative. The 2026-2027 research awardees1. Ensuring Intellectual Honesty Advancing LLM Reasoning Faithfulness without Faithfulness Rewards Faculty: Hao Peng Professor Peng is tackling the "hallucination" problem at its core. By developing methods to ensure large language models (LLMs) follow a logical, faithful reasoning path-without relying on traditional, often biased, reward systems-this work ensures that when an AI gives an answer, the "why" behind it is actually true, a critical component for trustworthy financial applications. TrACE-Contrast: Faithful & Consistent Code Reasoning via Trace-Aware Contrastive Learning Faculty: Talia Ringer and Reyhan Jabaarvand Writing code is one thing; understanding how it executes is another. Professor Ringer's project uses contrastive learning to align a model's code generation with its actual execution "trace." This ensures that AI-generated software is verified and logically sound, which is vital for maintaining the integrity of our core technology systems. 2. Bridging the Gap: Agency & Imagination Creative LLM Agents based on Thinking with Imagination Faculty: H ## Fragments: July 21 DevFeed: [Fragments: July 21]() Original publisher: [Read original article]() Author: Martin Fowler (martin@martinfowler.com) Published: 2026-07-21T13:13:00Z Content type: article Language: en Sources: [Martin Fowler]() Topics: [Vibe coding](), [Security](), [Large Language Model](), [Artificial Intelligence](), [Machine learning](), [Code generation](), [Software Engineering]() Tags: [ai](), [code-generation](), [data-quality](), [llms](), [modernization](), [security](), [testing](), [vibe-coding]() ### AI overview The article discusses findings from a software development retreat, emphasizing that verification has become more important as code generation improves. It examines risks from LLMs and vibe coding, including security concerns, poor contextual fit, weak testing, and inadequate data-quality controls. It argues for feedback sensors, organizational safeguards, and stronger engagement between engineers and executives. ### Source excerpt With this post, I'll wrap up my notes from the second Future of Software Development Retreat. But before I do, I should note that the full Thoughtworks report on the retreat is now available. They have five headline findings: Code generation is no longer the bottleneck -- verification is. 'Harness engineering' is emerging as a distinct, ownable discipline. Organizations are colliding with a real apprenticeship crisis. The executive/engineer expectation gap is a bigger risk than any technical limitation. Legacy modernization is the clearest, most defensible near-term value pool. ❄ ❄ A session convened around the mismatch of views about using LLMs between engineers using it and the C-suite and boards that were calling for it. The concern is that boards are looking at promised productivity gains, and not concerned enough about the risks, particularly about security. This was illustrated by one tale of a company that used ML-trained software to optimize the replacement of air filters on their field equipment. They were pleased to see that they were able to change the air filters less frequently, saving them $50 million. But the problem was the ML models were trained on equipment used in the desert, while their equipment was used in the arctic. Air filters in the desert deal with dust, but in the arctic the thing to remove is mosquitoes. There's an important difference here, mosquitoes rot, and enough decaying mosquitoes is a serious fire risk. Fires from such dead mosquitoes around infrequently replaced air filters cost the company $100 billion. Now such a tale could told of many situations without AI in the mix. Plenty of human situations have gone wrong when solutions are applied in a new context (which is why context is such a key word among pattern-writers). But the tale does remind us to be wary of an AI's suggestions, and to always think of how to build sensors to provide rapid feedback. Engineers particularly worry about the risks when citizen developers start vib ## Migrating from Apollo Tooling to GraphQL Codegen at Yelp DevFeed: [Migrating from Apollo Tooling to GraphQL Codegen at Yelp]() Original publisher: [Read original article]() Author: Igor Kusakov, Software Engineer Published: 2026-07-16T00:00:00Z Content type: article Language: en Sources: [Yelp]() Topics: [GraphQL](), [Code generation](), [migration](), [Tooling](), [React](), [TypeScript](), [monorepo]() Tags: [code-generation](), [deprecated](), [graphql](), [migrate](), [monorepo](), [react](), [tooling](), [typescript]() ### AI overview Yelp describes migrating its frontend React monorepo from deprecated Apollo Tooling to GraphQL Codegen for generating TypeScript definitions from GraphQL queries and the server schema. ### Source excerpt Introduction At Yelp, we rely heavily on GraphQL and Apollo for data loading in our frontend React monorepo. When a developer writes a GraphQL query or mutation inside a React component, the shape of the response is defined by the query itself -- but TypeScript has no way to know what that shape looks like at compile time. That's where code generation (codegen) comes in. We rely on codegen to read our queries and server schema, then produce TypeScript type definitions for every operation's input variables and output data. In our codebase, this looks like: src/components/UserProfile/ ├── UserProfile.tsx <- contains... ## What is the future of software engineering with Adam Bender, Principal Software Engineer at Google DevFeed: [What is the future of software engineering with Adam Bender, Principal Software Engineer at Google]() Original publisher: [Read original article]() Author: Dr Milan Milanović Published: 2026-07-02T15:01:06Z Content type: opinion Language: en Sources: [Tech World With Milan Newsletter]() Topics: [future of software](), [Software Engineering](), [AI-assisted coding](), [Testing](), [Integration testing](), [Google]() Tags: [ai](), [ai-coding](), [code-generation](), [conway-s-law](), [future-of-software](), [integration-testing](), [software-engineering]() ### AI overview The article discusses Adam Bender's view that the AI coding debate focuses too narrowly on speed. It argues that faster code generation increases pressure on testing, review, engineering culture, system understanding, and long-term maintainability, with integration testing becoming a particular challenge. ### Source excerpt Most of the AI coding debate is about speed. ## Creating an agentic feedback loop with reliability guardrails DevFeed: [Creating an agentic feedback loop with reliability guardrails]() Original publisher: [Read original article]() Author: Gavin Cahill Published: 2026-06-25T00:00:00Z Content type: article Language: en Sources: [Gremlin Blog]() Topics: [AI Development](), [AI-assisted coding](), [Chaos Engineering](), [Resilience](), [CI/CD](), [Code review]() Tags: [agentic-ai](), [ai](), [ai-for-code](), [chaos-engineering](), [ci-cd](), [code-generation](), [resilience](), [testing]() ### AI overview The article explains how reliability guardrails and resilience testing can create an agentic feedback loop for AI coding. It argues that agentic code review and QA alone may miss production failures, so fault injection and Chaos Engineering can validate system behavior under realistic failures and provide data that improves AI-generated code. It also discusses using resilience tests as an automated CI/CD gate. ### Source excerpt Reliability guardrails are essential for ensuring resilience with AI development, but they can also be used to create a feedback loop for AI context. ## Automated Schema Evolution in Pinterest's Next-Generation DB Ingestion Framework DevFeed: [Automated Schema Evolution in Pinterest's Next-Generation DB Ingestion Framework]() Original publisher: [Read original article]() Author: Pinterest Engineering Published: 2026-06-24T18:01:01Z Content type: article Language: en Sources: [Pinterest Engineering Blog - Medium]() Topics: [schema-evolution](), [Apache Iceberg](), [Automation](), [Databases](), [Apache Spark](), [Kafka](), [Pull Request](), [Code generation](), [datasets](), [monitor](), [Provisioning]() Tags: [automation](), [code-generation](), [datasets](), [engineering](), [icebergs](), [kafka](), [monitor](), [pinterest](), [process](), [schema-evolution](), [software](), [spark](), [storage]() ### AI overview Pinterest describes an automated schema-evolution framework for its CDC-based database ingestion platform, which uses Kafka, Flink, Spark, and Iceberg. The framework propagates supported schema changes across the stack, uses pull-request-based versioning and auditing, targets SLA-based consistency between online and offline schemas, and provides recovery paths for unsupported or ambiguous changes. ### Source excerpt Yisheng Zhou | Software Engineer II Liang Mou | Sr Staff Software Engineer Gabriel Raphael Garcia Montoya | Staff Software Engineer Istvan Podor | Staff Software Engineer Introduction In the first post of this series, we introduced Pinterest's next-generation CDC-based ingestion platform built on Kafka, Flink, Spark, and Iceberg. In production, upstream schemas are constantly evolving, and in a distributed CDC pipeline, schema is not just metadata -- it is a cross-system contract spanning ingestion, transformation, storage, and historical backfill. A schema change that is not handled carefully can break Flink jobs, block Spark upserts, or create inconsistencies between online and offline representations. This post walks through how we make schema evolution safe in practice: the onboarding model it builds on, the changes we support and the tradeoffs we accept, how updates propagate across the stack, and how we monitor and recover when things go wrong. We end with where we're heading next -- zero-gap schema evolution. Background & Motivation Schema evolution was particularly challenging in our environment because our ingestion pipeline is composed of multiple tightly related stages, each of which depends on schema in a different way. A single table onboarding flow spans CDC source configuration, Kafka provisioning, Flink and Spark code generation, Iceberg CDC and base table creation, and historical bootstrap -- all driven by the same schema. Once a source schema changes, that change must be reflected across code, metadata, and storage in a controlled and auditable way. Without a unified automation framework, manual updates across these layers increase the risk of drift, failed deployments, and inconsistent datasets. We therefore needed a way to make schema evolution safe, repeatable, and scalable. Our Solution: SLA-Based Automated Schema Evolution To address this problem, we built an automated schema evolution framework on top of our CDC ingestion platform, providing: au ## Type Level Security: The future of secure AI code generation? DevFeed: [Type Level Security: The future of secure AI code generation?]() Original publisher: [Read original article]() Author: Stephen Thoemmes Published: 2026-06-04T00:00:00Z Content type: article Language: en Sources: [Blog RSS Feed | Snyk]() Topics: [Security](), [AI-assisted coding](), [Vulnerabilities](), [Rust](), [Python](), [Web](), [Document Object Model (DOM)](), [Application Services]() Tags: [ai](), [application-security](), [awareness](), [blog](), [code-generation](), [code-security](), [developer](), [devops](), [devsecops](), [enablement](), [external](), [opa](), [python](), [rust](), [security](), [snyk-code](), [snyk-platform](), [tools](), [types](), [vulnerabilities](), [web]() ### AI overview The article argues that type systems and secure-by-design libraries can make many web application security vulnerabilities impossible to write or catch at compile time. It presents this approach as a way to reduce vulnerabilities in manually written and AI-generated code, with examples involving Rust, Python, and Trusted Types. ### Source excerpt Secure-by-design types can turn common bugs into compile-time errors. This post explores how type-level security could help prevent entire classes of AI-generated vulnerabilities. ## Introducing Mellum2: A 12B Mixture-of-Experts Model by JetBrains DevFeed: [Introducing Mellum2: A 12B Mixture-of-Experts Model by JetBrains]() Original publisher: [Read original article]() Author: Nikita Pavlichenko Published: 2026-06-01T15:45:17Z Content type: article Language: en Sources: [Hugging Face - Blog]() Topics: [Mixture of Experts (MoE)](), [Low-Latency Inference](), [Inference](), [Retrieval Augmented Generation (RAG)](), [Routing (disambiguation)](), [Benchmark](), [code-completion](), [hugging face]() Tags: [ai](), [benchmark](), [code](), [code-generation](), [coding](), [inference](), [latency](), [low-latency-inference](), [mixture-of-experts](), [model](), [rag](), [routing](), [software]() ### AI overview JetBrains introduces Mellum2, a 12B-parameter Mixture-of-Experts model trained from scratch on natural language and code. It activates 2.5B parameters per token to provide efficient, low-latency inference for routing, retrieval-augmented generation, summarization, sub-agent tasks, and high-throughput coding features. The model is released under the Apache 2.0 license and is reported to deliver competitive benchmark performance with more than twice the inference speed of similarly sized models. ### Source excerpt - Mellum2 is a 12B-parameter Mixture-of-Experts model trained from scratch on natural language and code. - The model activates only 2.5B parameters per token, making it efficient for high-throughput, low-latency inference. Mellum2 is can be used for routing, RAG, summarization, sub-agents, high-throughput coding features, and private deployments. - It is released under the Apache 2.0 license. ## Beyond code generation: rethinking engineering productivity in the age of AI agents DevFeed: [Beyond code generation: rethinking engineering productivity in the age of AI agents]() Original publisher: [Read original article]() Author: Yasmin McDowell,Lawrence Good,Ilya Yakovlev,Andrew Cheung,Binoy Dash,Simran Jumani,Dmitriy Meyerzon,Mark Breitenbach,Ishan Mishra,Kazuaki Okumura Published: 2026-05-28T18:00:00Z Content type: article Language: en Sources: [Dropbox Tech Blog]() Topics: [Artificial Intelligence](), [AI-assisted coding](), [code productivity](), [AI Development](), [Software Engineering](), [ci](), [pull-requests]() Tags: [agents](), [ai](), [ci](), [code-generation](), [code-quality](), [developer-productivity](), [developer-tools](), [developer-velocity](), [productivity](), [pull-requests](), [software-development](), [workflows]() ### AI overview Dropbox describes a shift from AI coding tools that accelerate implementation to agentic systems that execute scoped tasks. The article explains how increased code throughput creates pressure on review, CI, validation, release coordination, and production operations, requiring broader changes to engineering systems and workflows. ### Source excerpt How Dropbox is moving from AI tools that assist engineers to agentic systems that can execute scoped tasks, and how we're building platforms to support those workflows. ## Diverse reasoning traces teach LLMs to make better decisions DevFeed: [Diverse reasoning traces teach LLMs to make better decisions]() Original publisher: [Read original article]() Author: Sheng Jia; Xiao Wang; Shiva Kasiviswanathan Published: 2026-05-26T15:17:06Z Content type: article Language: en Sources: [Amazon Science homepage]() Topics: [Language models]() Tags: [benchmarks](), [chain-of-thought](), [code-generation](), [conversational-ai](), [deep-learning](), [evaluation](), [fine-tuning](), [iclr](), [iclr-2026](), [large-language-models-llms](), [llms](), [math-reasoning](), [parallel-reasoning](), [post-training](), [post-training-optimization](), [reasoning](), [reinforcement-learning](), [traces]() ### AI overview The article presents set-supervised fine tuning and global forking policy optimization to train LLMs on multiple distinct reasoning paths. It reports 5% to 7% single-shot accuracy gains on standard benchmarks. ### Source excerpt How to train language models to generate diverse, accurate reasoning paths using tokens that control distinct reasoning strategies. ## Promptimus: Improving already good LLM prompts with zero manual engineering DevFeed: [Promptimus: Improving already good LLM prompts with zero manual engineering]() Original publisher: [Read original article]() Author: Zhengyuan Shen; Yunfei Bai; Sullam Jeoung; Shuai Wang Published: 2026-05-14T13:47:45Z Content type: article Language: en Sources: [Amazon Science homepage]() Topics: [Automated prompt engineering](), [Prompt Engineering](), [Optimization](), [Large Language Model](), [AI Agent](), [Artificial Intelligence](), [multimodal](), [Code]() Tags: [ai](), [ai-agent](), [amazon-bedrock](), [automated-prompt-engineering](), [code-generation](), [conversational-ai](), [engineering](), [generative-ai](), [large-language-models-llms](), [llm](), [multimodal](), [optimization](), [performance](), [prompt-engineering](), [prompt-optimization]() ### AI overview Promptimus is an automated prompt-engineering method for improving already strong prompts without manual engineering. It uses task data, user-defined performance metrics, failure analysis, debugging agents, sanitization, and targeted edit mode to refine prompts while preserving working business logic. The method is model agnostic and supports textual and multimodal LLM tasks, including classification, extraction, generation, summarization, code generation, and tool use. ### Source excerpt By focusing on specific failure points and suggesting targeted solutions, a new automated prompt-engineering framework improves prompt performance without compromising existing functionality. ## Joining forces to modernize legacy software DevFeed: [Joining forces to modernize legacy software]() Original publisher: [Read original article]() Author: Mary Louise Stewart Published: 2026-05-14T08:00:00Z Content type: release Language: en Sources: [Federico Tomassetti]() Topics: [Software](), [Reverse Engineering](), [Refactoring](), [Code generation](), [interoperability](), [AI-assisted coding]() Tags: [code-generation](), [interoperability](), [legacy-code](), [legacy-modernization](), [migration](), [model-driven-development](), [modernization](), [parsing](), [refactoring](), [reverse-engineering](), [software](), [transpilers]() ### AI overview Strumenta and Metadev announced a collaboration to modernize legacy software. Strumenta contributes code analysis, extraction, and transformation capabilities, while Metadev contributes modeling, AI integration, and code generation. Their approach uses LionWeb models as an intermediate representation between legacy code and modernized systems. ### Source excerpt Strumenta and Metadev Join Forces to Modernize Legacy Software A collaboration combining Strumenta's expertise in code analysis, extraction, and transformation with Metadev's capabilities in modeling, artificial intelligence, and code generation from models. Torino/Seville, May 14, 2026 -- Strumenta S.r.l. and Metadev S.L. have reached a collaboration agreement to offer a joint approach to Modernizing Legacy [...] The post Joining forces to modernize legacy software appeared first on Federico Tomassetti. [Next page]()