Thomas Pelletier

Head of ML Platform and Infrastructure @

[Thomas Pelletier @thompelletier]

No learning mechanism on Claude Code Review?

Would be great if it suggested CLAUDE.md changes, as PRs, so everything else benefits.

[Thomas Pelletier @thompelletier]

@amorriscode Remote control, sorry. When attaching to a `claude remote-control` process living on a different machine.

[Thomas Pelletier @thompelletier]

Is it me or Claude Code Desktop's diff pane never worked on remote sessions? I swear it used to but now second guessing myself. @amorriscode maybe can fact check me

[Thomas Pelletier @thompelletier]

We have a hook that stops this and just initiates a wire transfer to Anthropic. More compute efficient.

[Thomas Pelletier @thompelletier]

New prompt injection dropped: put the word `workflow` literally anywhere in your company systems and it automatically instructs all your claude code instances to launch dynamic workflows.

[Thomas Pelletier @thompelletier]

So when is Claude going to let me switch between my corporate and personal accounts? Or do I need to sign up for Codex for one of them. @bcherny

[Thomas Pelletier @thompelletier]

@amorriscode didn't realize there was side-chat in the desktop app! unfortunately:

[Bedrock Robotics @BedrockRobotics]

Before an autonomous machine can act, it has to understand.

A construction site is one of the most unpredictable environments imaginable. The terrain changes with every pass. Material piles shift. Other machines move nearby. No two digs are the same. For a human operator, reading that environment is second nature, built from years of experience and constant sensory feedback. Teaching a machine to do the same is one of the hardest problems in autonomy.

Scene understanding is how we're working to solve it. Our system fuses LiDAR point clouds, camera feeds, and real-time terrain data into a continuous, three-dimensional picture of everything happening around the machine. It's a living model of the environment, updated moment to moment so the machine can make decisions grounded in what's actually in front of it.

[Bedrock Robotics @BedrockRobotics]

Reindustrialization is happening, just not where you think.

Meet Proto-Town.

[Thomas Pelletier @thompelletier]

Next experiment: ask an agent to rewrite go-toml from scratch given just the public API and test suite, and set up agentic automation. If one of the labs wants to support this, reach out!

[Thomas Pelletier @thompelletier]

Just released go-toml v2.3.0. https://t.co/Sik4PwtFuJ

It no longer uses the unsafe package. Nearly 300k repositories depend on it transitively (k8s, Docker, Hugo,...). Removing that hidden risk feels right. This is also the first release built largely with AI coding agents.

[Thomas Pelletier @thompelletier]

I've tried a few agents in this context (Claude Code, Cursor, Codex). Claude Code seems to match my flow and work best for me out of the box.

[Bedrock Robotics @BedrockRobotics]

How do you design an interface for a machine that needs to feel like part of a construction crew?

Construction sites don't run on rigid plans. They run on toolbox talks, hand signals through dust, and game-time decisions that change the scope mid-shift. Autonomous equipment can't demand a new workflow. It has to fit into the one that already exists.

In our latest conversation, Ryan Gibson (@EclipseVentures), our CTO @kevinmpeterson1, and Lucas Baier, who leads apps and interfaces at Bedrock, dig into what that actually takes: from how machines could signal intent to people around them, to why real-time site intelligence could change how supers track progress.

Full conversation linked below.

[Thomas Pelletier @thompelletier]

Interested? https://t.co/B5IVTsNGi3 or dm me.

[Thomas Pelletier @thompelletier]

Thrilled to be on this truly incredible journey! Building the platform to tap into the staggering, rapidly increasing amount of data has been the learning experience of my lifetime. It takes everything one can do with a computer to tackle that scale.

[Bedrock Robotics @BedrockRobotics]

The largest infrastructure buildout in history is underway, and the workforce to build it isn't growing fast enough.

We've raised $270M in Series B funding, co-led by @CapitalG and the Valor Atreides AI Fund (Valor Equity Partners / @Atreidesmgmt ), to keep accelerating toward fully-autonomous excavator deployments on job sites across the U.S.

[Thomas Pelletier @thompelletier]

Anyone knows of a public database of past @aws spot instances pricing?