Cole Lawrence @colelawrence.com · 20/11/2025I really loved sqlc's approach, so I'm glad there's another option like it for Rust! 020
Cole Lawrence @colelawrence.com · 16/11/2025I recommend Amp if you can afford passthrough pricing! I appreciate that they choose the best models for the job and I don't have to reconsider which models to switch to for different tasks every month 130
Cole Lawrence @colelawrence.com · 16/11/20254. If we enjoy working together through the trial and aligned your value to your desired compensation Only through the experience of a trial can we actually understand if the compensation you're seeking is aligned with your impact. And it confirms for you whether you enjoy working here with us. 020
Cole Lawrence @colelawrence.com · 16/11/20253. If we work well together Working together is natural and enjoyable as we work through an interesting technical design problem. You balance building with thoughtful API design and can talk through how it might work well or not for AI agents to maintain 110
Cole Lawrence @colelawrence.com · 16/11/20252. If we enjoy chatting about work It feels fun and effortless to chat. We can nerd out about similar things. I naturally learn something interesting through our conversation that I'll want to apply immediately. 110
Cole Lawrence @colelawrence.com · 16/11/20251. If you appear to be good. You are either a prolific product builder with experiences pivoting OR you have previously built something we need help building 110
Cole Lawrence @colelawrence.com · 16/11/2025How to pass each interview stage for the first 5 hires at our startup: 1. If you appear to be good => Intro 2. If we enjoy chatting about work experiences => Technical interview (pairing) 3. If we work well together => Trial period 4. If we align on impact<>desired comp => You're permanent 🙂 110
Reposted by Cole LawrenceRust NYC @rust.nyc · 30/10/2025First annual Rust NYC Unconf registrations are open! rust.nyc/unconf 0105
Reposted by Cole LawrenceNicolas Roman Posner @nrposner.com · 09/10/2025Finally wrote up my slides/notes for my presentation at the last #RustNYC meetup. Thanks to @busysp.bsky.social, @colelawrence.com, @statisticsftw.bsky.social, Ross, and the whole team over there for making it possible! nrposner.com/blog/rust-py... #rustlangnrposner.comRust-Python Interop for Scientific Codebases — nrposner 032
Reposted by Cole LawrenceNicolas Roman Posner @nrposner.com · 09/07/2025Announced the RustQuest community project at the @rust.nyc meetup yesterday! I hope this can grow as a resource to effectively teach young students #rustlang + CS concepts in tandem! special thanks to @colelawrence.com and @busysp.bsky.social for support! github.com/nrposner/rus...github.comGitHub - nrposner/rust_quest: An incremental, gamified introduction to Rust and computer science fundamentalsAn incremental, gamified introduction to Rust and computer science fundamentals - nrposner/rust_quest 1197
Cole Lawrence @colelawrence.com · 06/04/2025using LLMs to refactor and my new favorite prompt is becoming: "Ask me clarifying questions about any uncertainty you have about the refactoring." This is saving me so many little headaches where the LLM changed the approach in subtle ways. 030
Cole Lawrence @colelawrence.com · 30/03/2025I'm working through one of my toughest system designs ever—but I'm genuinely enjoying it. Usually, engineers steer the conversation too deep, while non-technical folks aren’t sure how to navigate the topics to ask great questions. 000
Cole Lawrence @colelawrence.com · 30/03/2025Absolutely killer combination for system architecture design: - Gemini 2.5 Pro for preparing interview questions - ChatGPT Voice mode for interview - Gemini 2.5 Pro for writing a technical write-up + updated overview instructions file for ChatGPT 120
Cole Lawrence @colelawrence.com · 18/03/2025This is our first month to have two events in a long time! Check it out if you're in NYC 031
Cole Lawrence @colelawrence.com · 08/03/2025Lovely primer on oklch, it feels like the cheat code to good color picking when you realize oklch and hct (material design) exist 010
Cole Lawrence @colelawrence.com · 07/03/2025Reasonable change. I have a pretty massive Svelte 5 codebase, but I think there are only a small handful of effect$ instances compared to state$ and derive$. 240
Cole Lawrence @colelawrence.com · 07/03/2025Most AI eng I see is usually just one step and done. Embed, store, retrieve, and synthesize. I want to see a portfolio where you have added steps in between that benefit the end user. Metaprompting, strategically using scratchpads, memories, tool loops with budgeting, etc 000
Cole Lawrence @colelawrence.com · 07/03/2025That would be taking novel approaches to solving problems with multiple steps. For example, show me how you wrote one AI to generate intermediary code to be run and used to scrape a website or organize data from multiple systems (this is something happening at Ramp when they import from new banks) 000
Cole Lawrence @colelawrence.com · 07/03/2025But, I would probably just end up abusing this for very little gain. There was a cool use case for this with typed link parameters for the Qwik router as well 000
Cole Lawrence @colelawrence.com · 07/03/2025Cool things are unlocked when TypeScript template literals become more capable of enforcing types. I'm interested in the inference as well, so you can retain the type based on what is written. For example, err`Permission not granted to ${user}` could be `Err<["Permission not granted to", User]>` 100
Cole Lawrence @colelawrence.com · 07/03/2025I learned about this from Stephen King's _On Writing_! 110
Cole Lawrence @colelawrence.com · 07/03/2025Yeah, that's just a different role at that point. If they come into he team not having a portfolio of AI experience with the intent to "train up" it's high risk to me that it doesn't click for them. Like hiring a production artist for a role in brand design 100
Cole Lawrence @colelawrence.com · 07/03/2025can you meet with a customer trying to accomplish a task that could be advised by those documents and design a pipeline that connect the dots to complement their workflow when the task they are accomplishing is completely unique to their IP? 100
Cole Lawrence @colelawrence.com · 07/03/2025given pdfs, calendar events, emails, and other docs with timestamps and associated people (who read it, authored it, shared it) where everything is related to 6 different complex projects at an organization... 200
Cole Lawrence @colelawrence.com · 07/03/2025Any tips on hiring AI product engs for a startup? Practical experience designing and troubleshooting tool-loops. $300k-$450k remote if they connect product thinking and AI pipeline setup, design, and devops experience. This is a hypothetical role at the moment, just looking for advice. 110
Cole Lawrence @colelawrence.com · 24/02/2025I started working on an open-source todo "UI" that syncs with a markdown file, since all my TODOs are in markdown for AI editing context. Also has a - Pomodoro timer - Customizable sound cues for break start, end, task completion, etc github.com/colelawrence... 000
Cole Lawrence @colelawrence.com · 05/02/2025Thanks for the explanation! Super interesting and simple approach to consider including immediately. Do you suspect that it would be effective to make a summary for each “Wait” gap to show “reasoning”? Or should you use a special tag/notation to make headers in the main completion? 120
Cole Lawrence @colelawrence.com · 04/02/2025Made a fun little helper for React to make it possible to "click" buttons without clicking when you have a input focused. Was having some trouble with debugging some focusout issue gist.github.com/colelawrence... 030
Cole Lawrence @colelawrence.com · 04/02/2025Music is either _the_ activity, and great speakers are amazing. Or, music is in the background and I easily forget how it sounds 010
Cole Lawrence @colelawrence.com · 18/01/2025I made something like this create.bingo a while ago, but for super simple things. I remember I was able to avoid configuration files almost completely by naming my variables with their types. E.g. {{typesDir}} or {{nameStrArr}}, then the cli args and prompts could be inferredcreate.bingoDelightful templates for web repositories. 💝create is a repository templating engine that allows you to instantiate ready-to-use repositories from flexible, powerful templates. 000
Cole Lawrence @colelawrence.com · 09/01/2025Also a good idea. I’m not sure how fast npx is at resolving where to execute the bin script 000
Cole Lawrence @colelawrence.com · 09/01/2025If you want to run the typescript directly, you could try node 23’s typescript support— it could certainly be faster and might not give you headaches with source mapping in your debugger nodejs.org/api/typescri...nodejs.orgModules: TypeScript | Node.js v23.6.0 Documentation 000
Cole Lawrence @colelawrence.com · 09/01/2025Not sure if it’s related, but if you are using a bloated version of zsh like oh-my-zsh and it’s set as the default shell, then that could add a second where bash is instant. 100
Cole Lawrence @colelawrence.com · 09/01/2025I need an agent that will just work on optimizing my typescript tsconfigs frfr 000
Cole Lawrence @colelawrence.com · 08/01/2025Sometimes I fall back into the habit of writing code with a keyboard, and I have to fight that habit formed from the last 10+ years For LLM coding to pay off, you gotta learn how to keep the context/session "on the rails", and getting "on the rails" is the hardest hurdle to get over 030
Cole Lawrence @colelawrence.com · 08/01/2025with AI, I seem to spend more of my time than ever designing the codebase's framework in a composable + testable way, because the payoff is huge when you know how each individual file is more complicated, but the total codebase is at a fixed complexity even with added features 020
Cole Lawrence @colelawrence.com · 06/01/2025I'm excited to be launching this new website for the next phase of @forethink.ai 🎉 I hope this page better communicates how we can amplify teams working in the browser by enriching the webpages they visit with private information from their company and peers–all with extreme care for data privacy!forethink.aiForethinkTransforming yours and your team's scattered data into enrichments on every page you visit. All data is private and flows directly between your computer and your trusted apps. 020
Cole Lawrence @colelawrence.com · 29/12/2024Great little write-up of today's best toolset for local LLMs Thanks for writing, Abishek ❤️ 000
Cole Lawrence @colelawrence.com · 29/12/2024Had the same experience yesterday when trying to watch it and not getting past the first 30 minutes 000
Cole Lawrence @colelawrence.com · 28/12/2024I used to have a ton of little USB chargers and I went through all of them and labeled their amperage because it was so annoying to not easily tell which chargers would be decent for charging vs continuous power applications 010
Cole Lawrence @colelawrence.com · 28/12/2024Oh yeah. That makes sense. I wasnt aware of that issue, since I was using a magsafe. Seems similar to when few knew that some USB chargers were slow at like 5V/300mA 010