hello, I'm ravens. I'm a "sr staff" software "engineer" for a bigco. I lead a team of ~30 product engineers, and help run an org of ~350.

I do a lot of other random stuff on the side.

you can find me here:

• @_R4V3N5_ on twitter
• ravens2d on github
• @rvns.moe on bluesky

no posts yet..

here are some projects I'm working on:

kawa - a fantasy retro console virtual machine, using a stack machine architecture with 16 bit words, 64k mem, 16k roms, and memory mapped IO

yuzu - a peer to peer chat protocol with permissionless network participation, global discovery, and pseudo-finality

ivy - a local first, p2p runtime using mDNS+DNS-SD for local network peer discovery, CBOR for serialization, ed25519 for peer identity, and a hand rolled diffie-hellman exchange for handshakes/wire encryption

yui - my personal claude in my terminal. supports episodic and factual memory. experimentation ground for other more fringe agentic features.

others you should check out, in no particular order:

• kennethnym
• hitorilabs
• vmfunc
• esp
• rose
• qtnx
• nearcyan
• vin
• echo4eva
• xilo
• dcpunks
• a11ce

• Active Data Curation Effectively Distills Large-Scale Multimodal Models
• DeepSeek-Prover: Advancing Theorem Proving in LLMs through Large-Scale Synthetic Data
• DeepSpeed Inference: Enabling Efficient Inference of Transformer Models at Unprecedented Scale
• GPipe: Easy Scaling with Micro-Batch Pipeline Parallelism
• Efficient Memory Management for Large Language Model Serving with PagedAttention
• Prover-Verifier Games improve legibility of LLM outputs

• Alpa: Automating Inter- and Intra-Operator Parallelism for Distributed Deep Learning

• Learn You a Haskell for Great Good!
• Category Theory For Programmers
• Keeping up with AGI (Google Doc)
• A (Long) Peek into Reinforcement Learning

• Understanding Multi-Head Latent Attention
• A vision researcher's guide to some RL stuff: PPO & GRPO
• You could have designed state of the art positional encoding
• How to Scale Your Model - A Systems View of LLMs on TPUs
• Mixture of Experts (MoE) - Deepseek-R1 Primer
• Multi-Headed Attention (MHA)
• DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
• Range-Based Set Reconciliation and Authenticated Set Representations
• Epidemic Broadcast Trees
• An Invitation to Deep Reinforcement Learning
• PyTorch Distributed: Experiences on Accelerating Data Parallel Training
• Scaling Distributed Machine Learning with the Parameter Server
• Delta State Replicated Data Types
• Sloppy hashing and self-organizing clusters (Coral Paper)
• HotStuff: BFT Consensus with Linearity and Responsiveness
• Cable Protocol
• Merkle Search Trees: Efficient State-Based CRDTs in Open Networks
• The Willow Protocol
• Database Internals: A Deep Dive into How Distributed Data Systems Work
• Beyond A*: Better Planning with Transformers via Search Dynamics Bootstrapping
• In Search of an Understandable Consensus Algorithm (The RAFT Paper)
• Ray v2 Architecture
• Designing Data Structures for Collaborative Apps
• The Grug Brained Developer - A layman's guide to thinking like the self-aware smol brained
• General-purpose CRDT libraries
• Automerge Libarary
• Software and Anarchy
• Peachcloud docs
• The Annotated GPT-2
• Building a GPT-like Model from Scratch with Detailed Theory and Code Implementation
• Scaling Instruction-Finetuned Language Models
• The Annotated Transformer
• You Want Modules, Not Microservices
• Earthstar docs
• Making CRDTs Byzantine Fault Tolerant
• netfarm source code / Netfarm Introduction
• p2panda docs
• Applied Language announces Applied Language
• Hyperhyperspace whitepaper
• Local-first software
• How Earthstar Works
• Scuttlebutt protocol guide
• Guide for new activitypub implementers
• cabal protocol
• DID Spec
• Rebooting the Web of Trust
• Web of Trust
• OrbitDB Docs
• GunDB Docs
• The Kademlia paper
• Hypercore Protocol