Tag
The tweet discusses using formal verification methods like TLA+ and Lean to ensure the correctness and scalability of mission-critical AI infrastructure software, with references to Intent Lab and Boris Cherny's work on the Claude Agent SDK.
The author describes an experiment where merging two AI agents' git worktrees led to test failures despite clean merges, highlighting the challenges of parallel agent development without mutual awareness.
SWE-Bench Pro V2 is an updated benchmark for evaluating AI agents in software engineering, featuring 642 tasks across 11 repositories with improved evaluation protocols and contamination controls.
Casey Muratori emphasizes the importance of learning assembly language for software optimization in a discussion shared by Gergely Orosz.
This article critiques the over-reliance on AI in software development, arguing that AI lacks wisdom for code maintainability and architecture, which can hinder developer expertise and lead to long-term issues.
A tweet sharing Lauren's talk on merging 2000 pull requests in a month, emphasizing creativity and practical workflow in software development.
A developer built a router to cut AI agent costs but found it only escalates requests, increasing spending; effective savings came from caching rather than routing.
Cognition introduces new CLI and SSH access for Devin Cloud, enabling developers to manage AI coding sessions directly from the terminal and interact with a dedicated VM.
A software engineer describes Claude as his cofounder, sparking a discussion on humanizing AI tools in software development.
Gergely Orosz is preparing a lecture for sophomore students on software engineering, revamping his 2024 talk with an angle emphasizing proactive learning.
This paper introduces SWE-Proof, a benchmark of formally verified code patches for real-world software issues, demonstrating that formal verification improves error detection in LLM-generated code and identifies specification synthesis as a key open problem.
A software engineer reports that AI models seem to be getting worse at following instructions in recent updates, often making unasked changes and ignoring contracts, leading to increased manual work.
Devin announces that SWE-2 is available for free in its Cloud Agents, CLI, and Desktop platforms until October 8 for all plans, encouraging users to try cloud agents at no cost.
This article compiles anecdotes from game developers detailing unconventional coding hacks and tricks used to overcome technical challenges and meet deadlines in game development.
StepFun introduces Step 5 Preview, a new flagship AI model for agentic work with frontier-level performance in software engineering and finance, featuring a 600B parameter scale.
This paper introduces a category-aware expert training framework for software engineering agents to mitigate uneven progress across task categories, using iterative training and multi-teacher distillation, with significant performance gains on Pro-618 and SWE-bench Multilingual benchmarks.
A developer explains building the open-source RE3 project for Mac, fixing mouse input bugs, and planning to add a B2 Stealth Bomber to GTA 3, focusing on software engineering methodology.
Devin, an autonomous AI software engineer, has launched macOS support in its cloud environment, involving a from-scratch rebuild of disk, networking, and provisioning systems in Rust to enable testing of macOS and iOS applications.
The blog post describes the 'senior engineer death spiral,' a cycle where engineers overwork and become isolated, leading to burnout. It offers advice on maintaining transparency and avoiding this pitfall.
This article summarizes Ion Stoica's talk at Ray Summit, exploring the three key gaps in requirements, environment, and evaluation faced by AI programming agents in software engineering, and how these issues lead to reward hacking and hallucinations.