 Conor Sweeney, MSCS

Being early isn’t always the golden ticket 🔑 I’ve seen plenty of Rust engineers take the leap into early-stage startups - some absolutely thrived, others burned out or watched the company fade before product-market fit ever arrived. But what separates the two? It usually comes down to leadership, focus, and whether the company is using Rust because it should, not just because it can. The truth is: joining early can be career-defining… or just chaotic. But when it’s the right team, the right product, and the right reason - the upside is unmatched. Have you ever joined a company too early, or got it just right? What made the difference for you...

7

Your legacy Express app is costing you your best engineers. 📉 I speak to Backend Leads every week. They all say the same thing. It’s not the salary that makes senior engineers quit. It’s the "Code Archaeology." 🦖 You know the feeling: • Digging through 5-year-old utils.js files with zero documentation. • Guessing what req.body actually contains because there are no types. • Being afraid to deploy on Fridays because test coverage is 12%. We often treat "Tech Debt" as a future problem. But in 2026, top-tier talent wants to work with modern tools (Fastify, TypeScript, Rust). Keeping a legacy Express monolith on life support isn't just "stable." It's a retention risk. We built Refactyl to automate the boring parts of modernization—so your team can focus on the fun parts (shipping features). Don't let your tech stack become a retirement home for bad code. Check your repo's migration score here: https://refactyl.com #techdebt #hiring #softwareengineering #developerretention #cto #refactoring

4

Six months ago, Spotnana engineering committed fully to Claude Code and Cursor. Adoption happened quickly and naturally. What surprised us most was a feature released last July in that many developers still haven't discovered: custom sub-agents. When you use ClaudeCode in its default form, you're working with a capable generalist. Custom sub-agents let you create specialists instead. You define them by placing a markdown file in the .claude/agents/ directory with simple YAML frontmatter that sets the agent's name, persona, and allowed tools. Claude then automatically delegates tasks to the appropriate specialist based on context and no manual prompting required. Each sub-agent runs in its own isolated context window, so your main conversation stays clean. The agents are also shareable across projects, which lets institutional knowledge build over time. The biggest challenge we faced was getting routing reliable. When agent descriptions overlap even slightly, Claude struggles to choose the right one. We've learned to make each persona description extremely precise and distinct. Another lesson came the hard way: sub-agents inherit your permission settings by default. I once gave a research-focused agent write access, and it began refactoring code while trying to investigate a bug. One improvement I would love to see is a clear post-session audit log that shows which agent handled which actions and when. Right now, tracing responsibility requires digging through long transcripts, which slows down debugging and makes accountability harder. If you're already using Claude Code (or Cursor with Claude) and haven't tried the /agents feature yet, you're missing out on significant capability. What sub-agents has your team built, and how have they changed your workflow? #ClaudeCode #AICoding #AgenticAI #CursorAI #DeveloperTools

60

3 Comments

I built sub-agents for Claude Code that are actually useful. The other night, my roommate Grant Glorioso asked what I thought of sub-agents in Claude Code. I told him most were too generic, just 'Backend Developer Agent' with a system prompt. They weren't solving a real problem. But then we started talking about context pollution. If Claude Code needs to go and find a single line of code, sometimes it has to read 20+ files, leaving it with a bunch of code in it's context window that is irrelevant to the problem we're trying to solve and fogs up its memory. Then we thought: What if sub-agents handled heavy operations in separate contexts and returned only condensed results? I sketched it on my whiteboard. Took a picture. Sent it to Claude. 30 minutes later, we were testing it on production codebases. The agents: house-research - Searches codebases/docs and returns condensed findings with file:line references instead of dumping full file contents into your context. house-bash - Runs commands and parses output for key info (pass/fail counts, relevant errors) instead of flooding your context with full logs. house-git - Analyzes diffs and commits, returns impact summaries grouped by severity instead of raw git output. Testing on real codebases, the token savings have been dramatic: 70k→3k, 43k→500, 21k→700. That's 95-98% reduction on heavy operations. Three more planned: house-mcp, house-vision, house-data These types of modern-day context engineering problems get me really excited about the future of AI and how to use it effectively. The bottleneck isn't the AI's intelligence anymore, it's how we architect our workflows around it. That's the problem space I want to be working in. I built it open source. Would love to hear if this resonates with your Claude Code workflow. Check it out here: https://lnkd.in/dmr_xwXQ

107

8 Comments

Robinhood just got its first underwriting deal, and it's easy to miss why that's a bigger deal than it sounds. ŌURA's IPO filing this week listed a surprise name in the underwriting syndicate: Robinhood Markets. It's the company's first official underwriting gig — ever. Here's the distinction that actually matters, and it's a product/architecture shift, not just a headline: Since IPO Access launched in 2021, Robinhood has only ever been a selling group member — investment banks price the deal, allocate shares, and Robinhood distributes whatever slice it's handed to retail customers. Passive downstream participant. Underwriter is a completely different role. Robinhood sits inside the syndicate that negotiates pricing directly with the issuer, sets deal terms, and takes on liability for the offering. That's the difference between being handed a fixed allocation and having a seat at the table deciding how big that allocation actually is. Why this matters for the product: Robinhood ended May with 27.7M funded customers — a retail demand pool large enough that companies going public increasingly can't ignore it. CEO Vlad Tenev has been explicit about the intent: use underwriter status to push for a materially larger share of IPO allocations going to retail, not just whatever legacy banks decide to spare. This follows Robinhood's underwriter approval back in June, timed right before SpaceX's Nasdaq debut, and sits alongside a broader pattern — the WonderFi acquisition, a reported $75M stake in OpenAI via private markets access, and a stated ambition to be "disruptive" in IPO structuring itself. The bigger picture: retail access to IPOs used to be entirely gatekept by Wall Street banks deciding what crumbs to pass down. Robinhood inserting itself into the syndicate — not just the distribution layer — is a structural bet that retail becomes a negotiated stakeholder in how IPOs get priced and allocated, not just a downstream customer of the decision. Worth watching whether other retail platforms follow, and whether legacy underwriters actually cede meaningful allocation share or just tolerate a symbolic seat. #Robinhood #FinTech #IPO #CapitalMarkets #Oura

5

Here's a software engineering question I hadn't really considered until now... Should programming languages evolve differently if AI is writing more of the code? Go 1.27 introduced several features that save developers time when writing code, but also add more abstraction to a language that was intentionally designed to be simple and readable. That's a pretty standard tradeoff for humans, but with AI now in the mix, things get a little more interesting. If agents are doing more of the writing, making code easier for humans to produce is going to become less of a priority. What we're going to care a lot more about is what the code looks like after it's written. Is it easy to understand? Easy to review? Can someone realistically maintain it? Those are the boxes we're looking to check. AI is obviously changing the way we write software. In fact, it's one of my favorite subjects to talk about on here. But I haven't really given a lot of thought to how it'll eventually change how we design programming languages themselves. Another fun thing to add to the list! Full article: https://lnkd.in/gzuJCx4i

4

5 Comments