Milo Bird

I've been surprised at how some simple improvements to our developer experience have improved our productivity so much. Some of our clients need to customise and change our simulations to fit their requirements; historically this has been difficult and often cost prohibitive, largely due to the amount of development time needed. Over the years we have built up a set of strong internal tooling to help us create and modify our products - we have a solid foundation that we build all of our simulations on and gives us a baseline of quality and reliability. However, this does not necessarily make it faster or more cost effective to build these products; largely it just improves their quality. Over the past few weeks I’ve been focusing on trying to improve our developer productivity, and I’ve been surprised at how much improved developer tooling and experience can affect things. Some of the changes we’ve made include: - Tightening the feedback loop, reducing time from code change to visible result - Stronger typing and validation, catching common mistakes and subtle errors at build time with clear error messages - Tooling for scaffolding code, rather than relying on copying things over from other projects we can use tools to set things up for us in a standard way - Code and type generation to allow us to check types end-to-end As a result of these changes I’m seeing: - Easier experimentation, the cost of trying something out is reduced - Errors caught earlier, saving debugging minutes which add up - More confidence in implementation, code is often correct on the first attempt - The cost of change has been reduced, as requirements change and we have to adapt that change is much less time intensive Another unexpected benefit is any AI tooling that we use is performing much better with these changes. AI is able to better validate its own work, spot issues faster, and correct them just as a developer would. This has been an extra productivity boost for us. Nothing I’ve been changing is particularly complex or novel, but the impact has been great. Often I think investment in developer experience and tooling can be neglected, primarily because it can be hard to quantify the impact of it, but I think the improved speed at which we can customise and build products for our clients is well worth it.

12

1 Comment

I built audio guide apps for museums using Bluetooth beacons. Sounds cool in theory. Reality was pure hellmo. Beacons work great in controlled environments. Museums are not controlled environments. Metal sculptures block signals. Glass cases create interference. Crowds of people with phones create chaos. Thick stone walls treat Bluetooth range estimates as fiction. I spent days climbing ladders, testing placement, getting yelled at by security guards for being too close to priceless art. "I'm just trying to get this beacon to work" was not a convincing explanation. The technical part was straightforward. Detect beacon, trigger audio, play content. Easy. The real world part had other plans. Beacons that worked perfectly during testing would randomly stop working when the museum opened. Signal strength varied by time of day, weather, and the number of smartwatches in the vicinity. I had to build backup plans for the backup plans. QR codes. Manual selection. GPS triggers. I learned that any technology depending on the physical world will break in ways you never imagined. Always have a plan B. And a plan C. The real world doesn't read documentation. What's been your latest dumpster fire disguised as a 'simple fix'?

6

6 Comments

“The developers who are most successful understand they’re going to be in a community…” Emilie Flanagan said that, and it reframed something I’ve seen for years but never quite put words to. A lot of teams still approach development like the hard part is getting the project sited correctly. But the real work starts when you have to defend your choices to the people who will live with them. That shift — from project to place — changes how you show up. And you can usually tell pretty quickly who gets that… and who doesn’t. 👇 What do you think developers still underestimate most about community engagement? Catch the Full Video SunCast Podcast

60

19 Comments

🥊 PTAL vs the Connectivity Metric: how do they compare? Two accessibility metrics. One city. Surprisingly different stories. At Podaris, we've been comparing London's PTAL scores against the DfT's new Connectivity Metric, and the results reveal something interesting key differences. 🚃 Take Croydon town centre: PTAL says it's a 6b (the highest possible score). The DfT Connectivity Metric? Closer to a 3. Same location, two very different answers. Planning applications in that area could justify car-free developments and no transport contributions due to the high PTAL score, but through the Connectivity Metric, should that be the case? They're both "right" - but they measure different things: 🚏 PTAL asks: "How easy is it to reach a bus or train?" 🏢 The DfT Connectivity Metric asks: "How many opportunities can you actually access?" Our statistical analysis of 500+ demographic variables shows the DfT metric better predicts real-world patterns - commuting behaviour, air quality, even deprivation indices. It correlates more strongly with 71% of the features we tested. But better prediction doesn't automatically mean better planning. In 800+ London neighbourhoods, these metrics disagree by three or more bands. That's 250,000 people living in areas where your choice of metric could determine parking requirements, density limits, or scheme viability. This isn't a bug. It's a feature of complexity. As accessibility analysis becomes more sophisticated -- accounting for destinations, travel times, mode choice, and behaviour patterns -- we gain richer insights. But we also introduce assumptions, data dependencies, and reproducibility challenges. The question for planners isn't "which metric wins?" It's: "How do we use both effectively while maintaining transparency?" 👇 Full analysis with maps, case studies, and a statistical deep-dive on here👇 https://lnkd.in/eesRyu6C #TransportPlanning #Accessibility #UrbanPlanning #PTAL #DataAnalysis #OpenData

284

18 Comments

Even while working on exactly this, I must say I'm surprised by a big brand like MANGO doing something exactly what ProdShot is building for all the small to medium shops out there. Reading the comments on the thread, understandably many in the industry are shook as the industry evolves. AI is here and moving fast, and it's affecting every industry out there, not even just software and digital but eventually hardware and the physical world as well. Many product photographers and designers feel uneasy about this, but to continue in the profession we all have to learn to adopt the new tools or risk falling behind. As a software engineer myself, I can tell you we go through the exact same thing. There's many of us who fully adopt the new world of using Cursor and other AI tools to develop, and some who adamantly refuse on principles. The fact is, the AI tools work, and they work really well. It doesn't mean you can vibe code your way to ship to production, and I can and have written many posts and threads about this: You have to use your software engineering and architecture knowledge and expertise to drive AI to develop the way you design your system. The lesson though is because it works so well, there will only be more and more companies and engineers who'll adopt it, before it becomes standard expectations in your profession. Instead of fighting it, learn how you can best use the tools and evolve together with the world.

3

I see so very many posts on here about AI changing the world, revolutionising everything, we're all going to be out of a job, you're a moron if you ignore it. Here's some brain noodles for you. AI, if you're a software engineer, is the single most well-read deep expert on every technology stack out there on your team. It's also the most inexperienced. It has little to no knowledge of your business, your legacy, why decisions were made a certain way, what trade-offs were argued over for hours in a meeting room. All it knows is that to solve problem X most people do Y. Treat it as such

21

"Forward Deployed Engineer" (FDE) is the hot new job title in AI. Never heard of it? That's because it's the fancy new rebranding of having a consultative arm in your company. This model has generally been looked down upon with the rise of product-led growth, but in the world of AI and enterprise implementations, it is increasingly the best way to succeed. While I don't love the buzzword, I do love the concept. I believe school districts want this too in their edtech partners (whether it's AI focused or not). They want more than just an out-of-the-box platform. They want a deeply hands-on implementation that aligns their long-standing processes with the technology.

14

Rare occasion of me posting something on LinkedIn full of experts and opinions, so I’m trying to share something genuinely useful. Over the last few months I’ve been very active in agentic engineering and kept running into tooling gaps I had to close for myself (and for a few people from the community). 1. Holding a local environment together with agentic tools is still annoying and doesn’t give enough control or flexibility. If I control execution myself, I waste time on terminal babysitting. If Cursor/Claude controls it, I lose visibility: I can’t properly read logs or understand what’s going on. So I built Crux, an AI-optimized local tool to run your app and keep it observable. The idea is simple: generate a config with an LLM, run crux in the terminal, and it starts your backend, frontend, mobile app, and dependencies, with logs you can actually follow. The main “magic” is built-in MCP support, so your agentic tool can also control the execution lifecycle: read logs, restart an instance, or trigger hot reload for a Flutter app. Tested on macOS only for now. If you make it work on Linux, contributions are welcome. https://lnkd.in/gnSAnWji [1/2]

6

Making 1-1s Actually Matter (Part 4): PDP Cookbook Plenty of IT companies genuinely care about their people – and many build real frameworks around self-development and career growth. I’ve seen engineers stay at the same company for years – not because they’re stuck, but because the path ahead is clear and motivating. When someone understands how their skills apply – and what they need to work on – that’s powerful. A clear growth roadmap isn’t just a perk. It’s a retention strategy. Each engineer sees things differently and as I mentioned earlier – working with people is never a trivial task. Can one framework fit everyone? Probably not. But that’s not a dead end. Not everyone’s chasing promotions. As a manager, know when to support and when to step back. But when growth is the goal, the PDP enters the game. A Personal Development Plan (PDP) is exactly what it sounds like – a roadmap of the key growth areas an engineer needs to focus on. I’ve heard the takes: - “We only use PDPs for junior folks”; - or the opposite: “Only mid-level and above need one”. In my view, a PDP can be useful at any stage. But like any tool, it only works when used right – and at the right time.  The first step? You need a clear snapshot of where someone is now. That baseline can come from their recent performance review, tech interview, or any structured assessment that reflects current skills. This is your starting point – not to judge, but to orient. Growth planning without it is like navigating without a map. Once you've identified the key development areas, the next step is simple – align them with the engineer. The best PDPs I’ve seen weren’t handed down – they were co-created. And that makes all the difference. When people contribute to shaping their plan, they own it. It’s a personal development plan for a reason. Let them set the deadlines too. In most cases, that works great. If they miss a target, they know the ball’s on their side – no need to micromanage. It builds ownership without pressure. And hey – sometimes, the engineer’s growth area is far from your own zone of expertise. That’s perfectly fine. We can’t know everything, can we 😀? This is where your internal network does the heavy lifting. If you’re not the right person to guide the topic hands-on, find someone who is. A strong PDP doesn’t mean you have all the answers – it means you know how to connect the dots. Pull in the right people. That’s how real growth happens. At the end of the day, close collaboration and goal alignment build a healthy, motivating environment.. When people understand their work connects to real growth and what steps they need to take – they’re more engaged, more focused, and more confident. It’s more than task completion. It’s about helping them see their value, map their future, and unlock the opportunities that come with it – from skill development to well-earned promotions. #Leadership #TeamManagement #CareerDevelopment #EffectiveCommunication #CompanyCulture

8

AI won’t replace senior mobile engineers. It’s exposing who has fundamentals. Over the last year I’ve leaned on AI for the boring bits—scaffolding tests, writing diff scripts, summarising crash threads. It’s great at speed. It’s terrible at judgement. Things I still have to do as a senior: Decide what not to build when battery, privacy, and deadlines clash. Catch the edge cases: killed-in-background, clock skew, flaky network, low RAM. Hold performance budgets (startup, jank, crash/ANR) and trade them like money. Ship safely: feature flags, staged rollouts, rollback plan, observability. Write docs people can follow at 2am. If I were starting today, I’d spend less time engineering prompts and more time mastering what AI can’t guarantee: State & lifecycles – predictable state, background work, cancellations, retries. Concurrency – structured concurrency/coroutines/async; race conditions and backpressure. Networking resilience – idempotent APIs, offline queues, conflict resolution. Performance – measure before tweaking: startup, memory, rendering; know your tools. Testing discipline – unit + integration + money-path UI tests; deflake strategy. Security & privacy – threat models, integrity checks, PII minimisation. Release engineering – one-click lane: bump → changelog → tag → upload → notify. UX empathy – accessibility, motion/haptics, empty/error states that don’t punish users. Use AI like a power tool: Generate boilerplate, test tables, and migration scripts. But always reproduce, benchmark, and verify before you ship. Users can’t tell if AI wrote a file. They can tell if your app starts fast, never loses data, and recovers gracefully. That’s craft. #MobileDev #iOS #Android #ReactNative #Flutter #AI #AppPerformance #CICD #Security #UX

17

1 Comment

🔥 Testing in the Age of AI 🔥 MCR.Test is back for 2026 on Thursday 12th March at Roku in Manchester! This isn't going to be another AI talk where someone shows you slides for an hour. We're actually doing stuff. 🙌🏼 Here's what we're covering: How to use AI to actually speed up your QA work (not just hype) Real examples of making testing and development smoother with AI Hands-on prompt injection testing — what it is, why it matters Interactive stuff where we break things and see what happens The point is — if you're doing QA or testing, AI is already changing how you work. We want to show you how to use it properly, not just talk about it. The details: 🗓 Thursday 12th March 2026 ⏰ 5:30PM till late 📍 Roku Offices, Circle Square, Manchester 🔗 https://lnkd.in/eF9cP9iN If you're in QA, testing, or automation and want to actually learn how to use AI in your day job, come along.  Cameron Whitwam Gina Chelton

13

3 Comments

It's not clear to me yet whether the current round of AI is just going to end up as fancy ruby on rails, good for scaffolding, or a really useful tool for dealing with large existing codebases. Certainly the agentic stuff I've played with so far seems to get bogged down pretty quickly. Has anyone summarised a good set of tools and techniques for tackling existing codebases?

2

1 Comment

What do all great developer tools have in common? Our founding engineer Matt just put out a new blog post with some thoughts around developer-friendly SDK design based around the last two years of iterating on the Hatchet SDKs. We treat our SDKs as part of our frontend, since they're usually the first entrypoint into Hatchet for a new user, and they're where our customers spend most of their time interacting with the product. To borrow an overused UX design term: We want using our SDKs to be a delightful experience. For us, "delightful" is based around a few principles: 1. Make it make sense - Can a developer understand what's happening without reading the manual? 2. Be opinionated to nudge developers towards the happy path - As developers build more complex systems, guide them toward patterns that will scale and succeed. 3. Teach along the way - Provide the resources developers need to grow. We finally got the Python SDK to the point where Matt will actually use it, which is a very high bar. Link to post in the comments ⬇️

21

1 Comment

How we’re hitting a 3-week time-to-hire for senior software engineers without cutting corners. Here’s a breakdown of how we filled a high-context senior frontend role at Web3 scaleup Rarible using the same system that’s now delivering 3-week TTH across our portfolio. Hiring high-bar engineers in Web3 isn't just about reach. It’s about rigour. We created an AI-enabled, go-to-market hiring campaign: The process (built for speed + quality) Clear ICP We mapped must-haves, dealbreakers, and likely objections up front, not after wasting 2 weeks. → Must-haves: React, TypeScript, Rx.js → No consultancies. Remote, GMT ±2. Web3/consumer background. Calibration table Turned the JD into a decision-making framework for screening, scorecards and syncs. → “Must-have vs. nice-to-have” clarity → Evidence > vibes → Helped the hiring team say yes/no faster, with less debate. Market mapping We broke down the talent pool into Tier 1 & Tier 2 across Europe. → 2,200+ profiles analysed → 96 Tier 1 engineers in Eastern Europe alone → Gave the client hard numbers + sourcing strategy options Structured interview kit → 10 categories → Scoring guidance for strong vs. weak answers → Reduced noise + sped up decision making Email outreach (that engineers actually respond to) → 4-part drip sequence (value prop, re-engage, product context, final nudge) → Personalised via GPT-4, signed off by the hiring lead → Tone: casual, confident, clear AI-enabled workflows → Automated outreach via n8n → Our Talent Partner operated like a full-stack squad Notion sourcing tracker → Per region: LinkedIn links, tags, message history → Fast async syncs with the client. No crossed wires. Prepped FAQ answers → Role, team, onboarding, tech stack, roadmap → Helped candidates convert Boolean done properly → Targeted for Rx.js + Web3 → Excluded low-context profiles → Localised by region The result ✔️ Fast, high-quality close ✔️ Happy hiring manager ✔️ Clean, repeatable system we now run across clients Hiring engineers in SaaS, Web3, fintech, or AI? This is the level of precision that gets results. DM me or connect if you want support.

66

11 Comments

Exciting news for developers using AppStoreConnect (at least haven't seen it elsewhere)! 🚀 The website has received a sleek UX update. Now, when attaching a build for review, the review page smoothly slides in from the right as a panel. This means no more getting navigated away - you can instantly move to review a bunch of apps without any interruptions. Streamlining the process for a more efficient workflow. #AppStoreConnect #DeveloperTools

7

1 Comment

🤯 Could X (formerly Twitter) be BANNED in the UK? Yep, you read that right. The UK government is considering a ban. 🔥Here’s the deal: It all comes down to "Grok," X's AI tool, and concerns over its misuse. 💡Grok is accused of being used to manipulate images, specifically "nudifying" photos of women and children. Think of it like an extreme version of a photo editing app gone rogue. 🚀 Ofcom, the UK's media regulator, is investigating. If they find X has crossed the line, the government is ready to pull the plug. What does this mean for you? * ✅ It highlights the urgent need for AI regulation. The Wild West days are over. * ✅ Every company integrating AI needs to prioritize ethical considerations *before* deployment. AI: innovation or a Pandora's Box? SOUND OFF below! This came from my daily auto-poster powered by n8n. I just build and let it post while I sleep 😴 #BuiltWithn8n #GrowthAutomation #Koddylabs

1

A combo of Gemma 4 26b MOE and Qwen 3.5 35b MOE is enough for 95% of what anyone needs an LLM for outside of non coding right now. Gemma 4 is stunning … Running absolutely at monster rates on a 5090, generating lyrics for Suno that are making me crease up. Frontier: * Sustained multimodal reasoning * The very top tier of complex coding/math * Massive context windows reliably What local now has: * Everything else * Privacy * No API costs * No content filters * Always on Qwen's tool calling is better. Gemma 4 though is even more a joy to talk to. It's ... GPT4o local.

3

3 Comments