A. Jesse Jiryu Davis is a Senior Staff Research Engineer at MongoDB specializing in distributed systems, Python concurrency, and serverless architecture. He contributes to MongoDB drivers, async frameworks, and writes on software design and open source projects.
51 articles from this blog
Analysis of CockroachDB's scalable leader lease protocol for multi-consensus groups, addressing failover and liveness issues.
Learn how to use TLA+ with LLMs to model and verify system correctness through a classic bean puzzle example.
Summary of the Antithesis BugBash Conference 2026, covering software verification, distributed systems, formal verification, and test coverage talks.
Review of a paper on using a meta-agent to automatically design novel AI agent architectures, outperforming hand-crafted systems.
Review of SysMoBench, a benchmark evaluating AI's ability to write TLA+ formal specs for real-world distributed systems.
Analysis of METR paper measuring AI's ability to complete long software tasks, showing LLM time horizons doubling every seven months.
Analyzes the decline of on-the-job training for junior engineers due to remote work and AI automation, posing a risk to the industry's future talent pool.
Explores how LLMs and AI challenge traditional definitions of privacy, analyzing the ethical implications of machine vs. human data observation.
Introducing a Zotero plugin to find and tag duplicate research papers by title or PDF hash, helping researchers clean up their libraries.
A computer scientist explains how large language models (LLMs) work, tracing their history from the Turing Test and ELIZA to modern AI, demystifying their operation.
Explores the impossibility of achieving common knowledge for coordinated attack in asynchronous distributed systems, based on Halpern and Moses' 1990 paper.
A critical review of a research paper proposing Raft optimizations for faster reads/writes while preserving linearizability in distributed systems.
Argues that YCSB and TPC benchmarks are flawed because they use closed-loop models, and advocates for open-loop benchmarks to better simulate real-world cloud database workloads.
Introducing LeaseGuard, a new lease protocol for Raft consensus that simplifies leader leases and enables faster crash recovery.
A guide to efficiently writing a Related Work section using Zotero and Overleaf, including tips for finding papers and managing citations.
A senior engineer shares his experience learning to code effectively with AI, from initial frustration to successful 'vibe-coding'.
A developer shares a personal productivity hack: using a browser bookmarks folder to aggregate and manage all their online feeds and sites.
A developer details their technical setup for displaying math and images across a blog, Atom feed, and email using Hugo and custom templates.
Explores the 'coordinated attack' problem in distributed systems, linking it to the impossibility of achieving common knowledge in asynchronous environments.
Explores the concepts of knowledge and common knowledge in distributed systems, starting with the classic muddy children puzzle.
