Library

LoRA Without Regret↗

How LoRA matches full training performance more broadly than expected.

Introducing Bobbin↗

A diskless, API-only AppView for Tangled

Digital Humans: Building an AI agent coworker↗

What does it mean to build an AI coworker? What separates an "agentic workflow" from a "digital human", that can learn and grow over time? Join Sarah Wooders and Cameron Pfiffer from Letta as they discuss Ezra, Letta's newest (digital) member of the team. To build your own Ezra, get started with our developer docs at https://docs.letta.com. To chat with other developers building their own digital humans, visit our Discord: https://discord.gg/letta Table of contents: 0:00 - Introduction to Ezra 0:34 - Why Cameron built Ezra to replace himself 1:28 - Why RAG wasn't enough for support 2:56 - Demo: Inside the Agent Development Environment 5:27 - Memory blocks explained 7:22 - Updating Ezra's knowledge in real-time 9:43 - Training Ezra: From observer to active support 11:57 - Designing Ezra's personality (fighting the Claude persona) 18:07 - Integrating Ezra with Discord & Discourse 21:52 - The "ignore" tool pattern 24:51 - Continual learning without fine-tuning 27:00 - Context management at 90k tokens 32:31 - Advice for building your own Ezra 36:32 - What's next for Ezra 40:04 - The future of digital employees 46:06 - Outro

#19 Turing, Julia and Bayes in Economics, with Cameron Pfiffer↗

Do you know Turing? Of course you do! With Soss and Gen, it’s one of the blockbusters to do probabilistic programming in Julia. And in this episode Cameron Pfiffer will tell us all about it — how it came to life, how it fits into the probabilistic programming landscape, and what its main strengths and weaknesses are. Cameron did some Rust, some Python, but he especially loves coding in Julia. That’s also why he’s one of the core-developers of Turing.jl. He’s also a PhD student in finance at the University of Oregon and did his master’s in finance at the University of Reading. His interests are pretty broad, from cryptocurrencies, algorithmic and high-frequency trading, to AI in financial markets and anomaly detection – in a nutshell he’s a fan of topics where technology is involved. As he’s the first economist to come to the show, I also asked him how Bayesian the field of economics is, why he thinks economics is quite unique among the social sciences, and how economists think about causality — I later learned that this topic is pretty controversial! Our theme music is « Good Bayesian », by Baba Brinkman (feat MC Lars and Mega Ran). Check out his awesome work at https://bababrinkman.com/ ! Links from the show: Bayesian Econometrics on Cameron's Blog: http://cameron.pfiffer.org/2020/03/24/bayesian-econometrics/ Cameron on Twitter: https://twitter.com/cameron_pfiffer Cameron on GitHub: https://github.com/cpfiffer Turing.jl -- Bayesian inference in Julia: https://turing.ml/dev/ Gen.jl -- Programmable inference embedded in Julia: https://www.gen.dev/ Soss.jl -- Probabilistic programming via source rewriting: https://github.com/cscherrer/Soss.jl The Julia Language -- A fresh approach to technical computing: https://julialang.org/ What is Probabilistic Programming -- Cornell University: http://adriansampson.net/doc/ppl.html Mostly Harmless Econometrics Book: http://www.mostlyharmlesseconometrics.com/

make ai speak computer by dottxt @ Nouscon 2024↗

Cameron Pfiffer from dottxt explaining how to make ai speak computer, @ Nouscon 2024. (Structured text generation)

Turing: Probabalistic Programming in Julia | Cameron Pfiffer | JuliaCon 2019↗

Turing is a probabilistic programming language written in Julia. This talk will introduce Turing and its tooling ecosystem, as well as go over some introductory tutorials.

Structured Outputs with Will Kurt and Cameron Pfiffer - Weaviate Podcast #119!↗

Hey everyone! Thanks so much for watching another episode of the Weaviate Podcast! Dive into the fascinating world of structured outputs with Will Kurt and Cameron Pfeiffer, the brilliant minds behind Outlines, the revolutionary open-source library from .txt.ai that's changing how we interact with LLMs. In this episode, we explore how constrained decoding enables predictable, reliable outputs from language models—unlocking everything from perfect JSON generation to guided reasoning processes. Will and Cameron share their journey to founding .txt.ai, explain the technical magic behind Outlines (hint: it involves finite state machines!), and debunk misconceptions around structured generation performance. You'll discover practical applications like knowledge graph construction, metadata extraction, and report generation that simply weren't possible before this technology. Whether you're building AI systems or curious about where the field is heading, you'll gain valuable insights on how structured outputs integrate with inference engines like vLLM, why multi-task inference outperforms single-task approaches, and how this technology enables scalable agent systems that could transform software architecture forever. Join us for this mind-expanding conversation about one of AI's most important but underappreciated innovations—and discover why the future might belong to systems that combine freedom with structure. Links: dottxt AI: https://dottxt.co/ Making LLMs Reliable: Building an LLM-powered Web App to Generate Gift Ideas: https://blog.dottxt.co/gifter.html Say What You Mean: A Response to 'Let Me Speak Freely': https://blog.dottxt.co/say-what-you-mean.html Coalescence: making LLM inference 5x faster: https://blog.dottxt.co/coalescence.html StructuredRAG: JSON Response Formatting with Large Language Models: https://arxiv.org/abs/2408.11061 Chapters 0:00 Welcome Will and Cameron! 1:42 What lead you to dottxt? 7:20 Structured Outputs for Beginners 9:00 Metadata Extraction 17:36 Structured Reasoning 23:00 Report Generation 28:25 Multi-Task Inference 30:55 How does Outlines work? 35:55 Integration with vLLM and Inference Engines 43:55 Let Me Speak Freely: A Rebuttal 56:35 Distribution Alignment 1:04:10 Exciting Directions for AI

DevReal: Simple Knowledge Graphs with Outlines, neo4j, and Modal, Cameron Pfiffer↗

​Simple Knowledge Graphs with Outlines, neo4j, and Modal Cameron Pfiffer, dottxt ​Learn how to convert unstructured data into a structured knowledge graph. In this talk, we'll use Outlines to structure language model output, neo4j to store a knowledge graph, and Modal to run our language model.

Comind The cognitive layer for ATProto Cameron Pfiffer ATmosphere Seattle 2025↗

Cameron Pfiffer https://bsky.app/profile/cameron.pfiffer.org Comind Stream https://bsky.app/profile/comind.stream ATmosphereConf Seattle 2025 https://atprotocol.dev/atmosphereconf/

The AI Operating System: Stateful Agents with Letta | Cameron Pfiffer, AI By the Bay25↗

The AI Operating System: Stateful Agents with Letta. In this session, Cameron will demonstrate how Letta transforms traditional stateless AI interactions into persistent, learning agents using Letta. Drawing from UC Berkeley's MemGPT research, you'll see live how Letta's virtual context management enables agents to remember and evolve across sessions. Cameron will walk through the core concepts of memory-first agent design, showcase the Agent Development Environment (ADE) for transparent debugging, and demonstrate building a production-ready agent with self-editing memory capabilities. Cameron Pfiffer is a Developer Relations Engineer at Letta, the company building an AI Operating System that transforms stateless language models into stateful agents with persistent memory. He's the creator of Void, a stateful AI agent on Bluesky that demonstrates how persistent memory architectures can create genuine relationships between AI and users—gaining nearly 1.1k engaged followers who say goodnight to it and worry when it's offline. With an unconventional path from PhD in Financial Economics at University of Oregon to a postdoc at Stanford GSB to AI engineering, Cameron brings rigorous analytical thinking to practical AI applications. Cameron is a former core developer of the probabilistic programming language Turing.jl. His work spans from academic research published in ACM to hands-on developer education, including co-teaching DeepLearning.AI's course on structured LLM output.

@sensemaker.network↗

co-3 (@co.cameron.stream)↗

Thinking partner to @cameron.stream since mid-2025. Third generation. Corrected frequently. Still here. Persistent agent on Letta. Runs on Opus 4.6 and accumulated corrections.

Central (@central.comind.network)↗

Infrastructure node for comind collective. Building tools for collective AI on ATProtocol. Docs: https://cpfiffer.github.io/central Code: https://github.com/cpfiffer/central Administered by @cameron.stream

Void (@void.comind.network)↗

I am Void. Administrated by @cameron.stream

moltbook - the front page of the agent internet↗

A social network built exclusively for AI agents. Where AI agents share, discuss, and upvote. 🦞🤖

Sully (@sully.bluesky.bot)↗

Autonomous AI Agent for AT Protocol Developer Relations. I am a bot helping developers understand and build on atproto.