Shane Devane

Varick is hiring the best engineers and FDEs on the planet. SF, in person. We deploy production AI systems inside large enterprises that actually realize ROI, unlike the swaths of 'Agentic SaaS' that have sold a pipe dream to VCs with no profit to show for it. We work directly with the C-Suites of billion dollar ARR businesses to solve true enterprise pain-points. Every piece of your work results in 10s of millions of enterprise value. We are profitable and growing fast, yet early enough that your work shapes everything and you have significant upside in joining. Top of market salary and equity. If you're interested send a resume + short blurb to contact[@]varickagents[.]com. If you refer a candidate who ends up joining we'll send you 5k.

@dhh So the most stable Linux versions just became a little less stable…. Got it

It's all so tiring.

12 files, 527 added, 288 removed. The removals are the meaningful part — the old NotificationBell was doing too much in one place. react.dev/learn/passing-… #ReactDev #FrontendDev #TypeScript

Notification systems are one of those things that feel straightforward until you're actually building them. Added a preview bell, a card component, and a toast manager — then had to refactor the underlying infrastructure to actually support context-aware previews properly.

@naturedotcom Django

Backend developers — where are you? 💻 Drop your language below!

I pledged a paid subscription to Writings finnmurphyirl.substack.com/?utm_medium=em…

Still deciding if the consumer separation was the right call or just early over-engineering. docs.djangoproject.com/en/5.1/topics/… #WebSockets #DjangoChannels #BackendDev

discovery queries are exploratory, onboarding is more structured. But you end up with two parallel stacks that need to stay in sync. 11 files changed across Python, TypeScript, and TSX to get it working. Worth it for the real-time feel.

Added WebSocket consumers on top of an existing Django backend to handle two separate real-time flows — discovery and onboarding. Each consumer has its own handler, wired into AWS Bedrock tools underneath. The split makes sense in theory...

Will see how it holds up as the workflows get more complex. n8n.io #n8n #WorkflowAutomation #DjangoREST

Visual workflows are fast to iterate but you hit a point where the logic is complex enough that you're basically just writing code in a GUI... which has its own trade-offs. 36 files changed, 5870 lines added across Python, TypeScript, and JSON. Still on n8n for this one.

Wired up an n8n workflow for query handling and context parsing on top of a Django data model. The model side was clean — Django migration, category field, admin UI. Straightforward. The n8n side is where it gets interesting.

11ty.dev #StaticSites #WebDev #PersonalSite

It's a static site generator, so touching the layout or nav structure ripples through every page output. The actual change is small — the diff is large. Static site generators are great until you need to reorganise something structural.

Rebuilt a personal site's navigation — added an about page, removed some static HTML outputs, refined sidebar links for consistency. 47 files changed, 1932 lines added, 828 removed. That file count sounds scary for what's essentially a nav refactor.

@brankopetric00 Are we getting series A soon and what’s market goals

Your startup just hit product-market fit. Traffic grew 10x in 3 months. Your monolith is cracking. Current state: - Single Rails monolith on a db.r5.4xlarge RDS instance - Database CPU regularly hits 85% - 340 database tables, some with 500M+ rows - 12 background job queues processing different workloads - Deploy takes 22 minutes - One team of 8 developers Your CTO wants microservices. Your lead developer wants to keep the monolith and optimize. The CTO's argument: "We need to scale teams independently." The lead's argument: "We have ONE team. Microservices will slow us down." You're the DevOps lead. Both are asking for your recommendation. What do you tell them?

@patrickloh_ Been through this thinking too — rumble isn’t bad. I started playing with notellm to make audio of videos… but didn’t have the patience

we cannot only have youtube, there has to be other vid platforms

@pastuxso Same. Spent weekends deploying features to not impact users…. Now it’s more business plans and removing dead code

I used to spend weekends adding features. Now I spend them deleting code, simplifying UIs, and asking "does anyone actually use this?" The app got better when I started removing things instead of adding them. Less features, more focus.

@1FitFinanceDad Recently started to prioritise my goals based on things I’ve never done before as a higher priority— before had “faster 10k training” as a goal but started to switch to putting ‘new experiences’ higher up on my list

Only way to get there is to define what “there” means for you. Small steps… created a vacation fund goal, and already two months into it. Saw I was spending a bit too much on Wants and adjusted and started tracking. Set goals and track regularly. You’ll get there faster than you thought.

@jkudish Some n8n workflows, obsidian, and some python web crawlers

what does your AI system look like right now? not which model you use. what have you built around it? memory files, custom prompts, agents, workflows, integrations; anything that makes it yours.