تسجيل الدخول لعرض الملف الشخصي الكامل لـ Matt
أو
عضو جديد على LinkedIn؟ انضم الآن
بالنقر على الاستمرار للانضمام أو تسجيل الدخول، أنت توافق على اتفاقية المستخدم واتفاقية الخصوصية وسياسة ملفات تعريف الارتباط على LinkedIn.
تسجيل الدخول لعرض الملف الشخصي الكامل لـ Matt
أو
عضو جديد على LinkedIn؟ انضم الآن
بالنقر على الاستمرار للانضمام أو تسجيل الدخول، أنت توافق على اتفاقية المستخدم واتفاقية الخصوصية وسياسة ملفات تعريف الارتباط على LinkedIn.
مانشيستر، المملكة المتحدة
تسجيل الدخول لعرض الملف الشخصي الكامل لـ Matt
يمكن أن يقدمك Matt إلى أكثر من 10 من الأشخاص في Nearform
أو
عضو جديد على LinkedIn؟ انضم الآن
بالنقر على الاستمرار للانضمام أو تسجيل الدخول، أنت توافق على اتفاقية المستخدم واتفاقية الخصوصية وسياسة ملفات تعريف الارتباط على LinkedIn.
٦١٣ متابع
٤٧٥ زميل
تسجيل الدخول لعرض الملف الشخصي الكامل لـ Matt
أو
عضو جديد على LinkedIn؟ انضم الآن
بالنقر على الاستمرار للانضمام أو تسجيل الدخول، أنت توافق على اتفاقية المستخدم واتفاقية الخصوصية وسياسة ملفات تعريف الارتباط على LinkedIn.
عرض الزملاء المشتركين مع Matt
يمكن أن يقدمك Matt إلى أكثر من 10 من الأشخاص في Nearform
أو
عضو جديد على LinkedIn؟ انضم الآن
بالنقر على الاستمرار للانضمام أو تسجيل الدخول، أنت توافق على اتفاقية المستخدم واتفاقية الخصوصية وسياسة ملفات تعريف الارتباط على LinkedIn.
عرض الزملاء المشتركين مع Matt
أو
عضو جديد على LinkedIn؟ انضم الآن
بالنقر على الاستمرار للانضمام أو تسجيل الدخول، أنت توافق على اتفاقية المستخدم واتفاقية الخصوصية وسياسة ملفات تعريف الارتباط على LinkedIn.
تسجيل الدخول لعرض الملف الشخصي الكامل لـ Matt
أو
عضو جديد على LinkedIn؟ انضم الآن
بالنقر على الاستمرار للانضمام أو تسجيل الدخول، أنت توافق على اتفاقية المستخدم واتفاقية الخصوصية وسياسة ملفات تعريف الارتباط على LinkedIn.
نبذة عني
مرحبًا بعودتك
بالنقر على الاستمرار للانضمام أو تسجيل الدخول، أنت توافق على اتفاقية المستخدم واتفاقية الخصوصية وسياسة ملفات تعريف الارتباط على LinkedIn.
عضو جديد على LinkedIn؟ انضم الآن
النشاط
٦١٣ متابع
-
Matt Harrison شارك ذلكView my verified achievement from Anthropic.Claude Certified Architect - Professional was issued by Anthropic to Matthew Thomas Harrison.Claude Certified Architect - Professional was issued by Anthropic to Matthew Thomas Harrison.
-
Matt Harrison أعاد نشر ذلكMatt Harrison أعاد نشر ذلكDax (co-creator of OpenCode) with yet another banger on the reality of LLMs in the actual workforce. All his claims are backed by research, by the way: https://lnkd.in/eSsFqrS2
-
Matt Harrison شارك ذلكI am somewhat embarrassed to admit that I reached halfway through my master's degree in AI/ML without truly grasping the bias-variance tradeoff beyond an abstract level. The dartboard analogy never really helped either. It finally clicked when looking at learning curves, and I decided to make a poster (yes, with AI). Hopefully someone else finds it useful too.
-
Matt Harrison نشر ذلكThere's a lot of feelings in Software Engineering right now. Excitement. Fear. Incredulity. Sometimes all at once. I think where you sit on that spectrum depends on why you got into this field in the first place. For me, it was never about the job title, or the potential salary. The first thing I built was a Flash soundboard. The kick I got from having an idea in my head, seeing it on the screen, and showing my friends - I was hooked from that moment. Is my job safe? Honestly, I don't know. Will the profession change? Undoubtedly. But I still have ideas. I still want to see them exist. I still want to show people. And I've never been able to do that faster. So if you're feeling uncertain right now, remember why you started.
-
Matt Harrison نشر ذلكI'm trying a new way of reading research papers. I often skim them to get the most important insights and gloss over some unfamiliar terms or scary looking formulas. Or sometimes I use the SQ3R method, where I write down questions and answer them through reading. I think there's still value in that, however getting a really deep understanding is infinitely more rewarding. My new approach is an extreme form of Deep Reading; the single rule is you can only read the next sentence when you fully grasp the current one with no holes in your understanding. This is very slow and painstaking and can have a high branching factor because each question can mean you need to read several other sources first. Working with an LLM can help to prune this tree though by comprehensively resolving tangential questions in one place, meaning you don't sacrifice depth and keep moving.
-
Matt Harrison نشر ذلكI need to admit something difficult...I'm guilty. I've been too dismissive of Vibe Coding. Non-developers are talking about how this makes everyone a developer now, maybe we won't even need any engineers soon. Meanwhile, experienced engineers argue that Vibe Coding is like taking the fast lane to technical debt, creating a hairball of code with no consideration for architecture, and letting all the non-functional requirements (maintainability, scalability, security etc) go out the window. On reflection, I think they're actually both right. However if we stop there, we're missing out on an a big opportunity to improve how we build products. A major bottleneck in product teams is the disconnect between what users need and what engineers build. Vibe Coding, i.e tools that let people "build" by describing intent rather than writing code blurs the line between ideation and implementation. Instead of either blindly embracing Vibe Coding to produce our products or dismissing it outright, we should focus on how we can integrate this into the development lifecycle thoughtfully. Given that a large % of your ideas aren't going to actually solve your users needs, we need a way to quickly evaluate ideas with minimal cost. These tools let customer experts, designers, and product managers go from idea to prototype in minutes, not weeks, no waiting on a ticket in your backlog to be groomed to "ready". But here's the critical thing: prototypes aren’t products. We still need engineers to turn validated ideas into robust systems. The challenge then is to figure out where Vibe Coding ends and engineering begins and how to smoothly transition between the two. Imagine a workflow where: - Non-engineers prototype with Vibe Coding tools to explore ideas and gather early feedback - Once an idea shows promise, engineers step in to validate and re-architect it with the right patterns, performance considerations, and non-functional requirements. - Product and engineering collaborate closely, informed by data instead of assumptions about what users need If we can get this right, we can spend less time building the wrong thing, and more time building the right thing right. And nobody worries about being replaced by robots.
-
Matt Harrison أعاد نشر ذلكMatt Harrison أعاد نشر ذلكI've spent about four hundreds hours testing "vibe coding" (or what I used to call conversational programming) to the extremes, and I've been telling people that it's great but not useful for anything other than prototypes. For this, I've got a lot of earache from some saying that I just don't understand it. The guy who created the term "vibe coding" finally bothers to try and build something end to end with vibe coding - "MenuGen is my first end-to-end vibe coded app" - https://lnkd.in/e9PqWwv6 - and comes to a conclusion that "Ultimately, vibe coding full web apps today is kind of messy and not a good idea for anything of actual importance." - https://lnkd.in/e9PqWwv6 Nothing wrong with the conclusion but, hang on - for the first time! Did the entire industry lose its mind over a term created by a person with no actual experience of doing what the term means!?!?? Talk about cult and a cult of personalities. Nothing surprises me in the world of AI these days. Let us hope that Gene Kim and Steve Yegge have actually tried to build entire applications before they finish the book - https://lnkd.in/eer2Pcb4 Vibe coding is a really poor choice of name. This is more wrangling than coding. Yes, people often confuse it with software engineering + AI (where a person inspects and reviews the code) but that's a different field to one where "I wrote none of the code directly (Cursor+Claude/o3 did) and I don't really know how the app works, in the conventional sense that I'm used to as an engineer". The confusion stems from the use of the word "coding" (hence my preference for wrangling). Technically prompts are code, they are just non deterministic and result in deterministic code. Hence I understand why people have used "coding" along with the vibe term. An old friend of mine Simon Willison nailed the confusion with this post - https://lnkd.in/ewXWwdX7 “generating code with AI without caring about the code that is produced” is vibing. Well, technically it's more "generating code with AI without being involved in the decision making" as we discuss in rewilding software engineering - https://lnkd.in/ek9MF7P3 Vibing will be important and bizarrely enough ... it is a skill. I'll write more on that later, I just need a few more hundred hours of experience to nail down the practices or at least to be able to have a good first stab at them. Even then, take them with a pinch of salt - we are still learning as the practices emerge. #vibing #vibe #ai #cult -- addendum X : Will it ever be useful for more than just prototypes? Me : Of course. For now though, we haven't developed the practices and we have to rely on software engineering i.e. reviewing the code. Those practices are emerging and the boundary of where to vibe (or not) will change over time. But it will take time. X : This year? Me : You've been listening to Eric Schmidt? Now that's a person I can almost guarantee hasn't used this stuff in anger but still pontificates about it.
-
Matt Harrison نشر ذلك"You don't need to prepare anything, just speak off the cuff." People say this because they think it makes life easier, but for some, myself included, the idea of speaking on a topic coherently with no preparation is akin to baking a soufflé blindfolded: chaotic, nerve‑wracking, and almost guaranteed to be a flop.
-
Matt Harrison أضاف إعجابًا إلى ذلكMatt Harrison أضاف إعجابًا إلى ذلكWe have a new website. Built natively in Next.js and Three.js. Because we don't do shortcuts and workarounds. We raised our pre-seed earlier this year. There's been a lot of changes since then. A new website is perhaps the most obvious. If you are on the list for access to PRISM, or signed up to our Networks Activity Summaries, this will be your portal. Get in touch if you want to know more about Valent, sign up to our summaries or anything else.
-
Matt Harrison أضاف إعجابًا إلى ذلكMatt Harrison أضاف إعجابًا إلى ذلك2026 marks my 10th year at Nearform, as well as my 45th birthday. I find the symmetry somewhat fascinating, and I've decided to honor it by taking a break from Nearform, software engineering, and work altogether. Staying would have been the easy choice. In ten more years I'd probably have more experience, a higher role, and considerably more wealth. But I would also be 10 years older. I wrote down some thoughts that don't fit in a post. They are in the article below.
-
Matt Harrison أضاف إعجابًا إلى ذلكMatt Harrison أضاف إعجابًا إلى ذلكQuite happy with this little side quest - check out the free version here, featuring occasional industry commentary: https://lnkd.in/dP3Bzmxt
-
Matt Harrison تفاعل مع ذلكMatt Harrison تفاعل مع ذلكI've found my nerd tribe! Having loads of fun and already digging many rabbit holes 🕳️ 🐇 as I joined Nearform 🤓 Just a happy face Friday photo, sweet humans. Much love 💚 #convolutedM #languagenerd #EngineeredDifferently #AINativeEngineering #AIAgents #LLMS
-
Matt Harrison أضاف إعجابًا إلى ذلكMatt Harrison أضاف إعجابًا إلى ذلكThe fact that Agatha Christie had access to ChatGPT is undeniable proof that time travel is real ________ photo credit: fio D.
-
Matt Harrison أضاف إعجابًا إلى ذلكMatt Harrison أضاف إعجابًا إلى ذلكI’ve got three kids, so a work life balance is non-negotiable. (Just ask my wife!) 👀 However, getting that balance right did not happen quickly. It took years of getting it wrong, a couple of tough periods, and some honest reflection before I figured out what actually worked for me and my family. Now we're pproaching 15 years in business, we've moved past the times where I need to be so 'on'. My focus is mostly on looking after our clients and overseeing our wider growth - but I've been stepping back to allow my team to shine. This presents better opportunities for them to grow, and gives me and my family the time we need. Hallelujah! 🙌 Any other parents and business owners in my network, how are you finding striking this balance currently? #mobileapps #appagency #business
-
Matt Harrison أضاف إعجابًا إلى ذلكMatt Harrison أضاف إعجابًا إلى ذلكRemember when we used to care about comma first?
-
Matt Harrison أضاف إعجابًا إلى ذلكMatt Harrison أضاف إعجابًا إلى ذلكAI did everything but discover a bug I was searching for today. You know what found the bug in the end? console.log and a pair of eyes. AI is a fantastic tool but it will be a while before humans are out of the loop.
الخبرة والتعليم
-
NearForm
** ********* *********
-
***********
********** ********* ************
-
****** *******
****** *********** *********
-
******* ********* ** **********
****** ** ******* * ** ********* ******** ******* ******** ******** *************** undefined
- الحالي
-
*** ********** ** **********
******** ** **** * ** ************
-
عرض خبرة Matt الكاملة
تعرّف على المسمى الوظيفي للأشخاص ومعدل بقائهم في العمل والكثير غير ذلك.
مرحبًا بعودتك
بالنقر على الاستمرار للانضمام أو تسجيل الدخول، أنت توافق على اتفاقية المستخدم واتفاقية الخصوصية وسياسة ملفات تعريف الارتباط على LinkedIn.
عضو جديد على LinkedIn؟ انضم الآن
أو
بالنقر على الاستمرار للانضمام أو تسجيل الدخول، أنت توافق على اتفاقية المستخدم واتفاقية الخصوصية وسياسة ملفات تعريف الارتباط على LinkedIn.
التراخيص والشهادات
المنشورات
-
hapi.js in Action
Manning Publications
عرض المنشورhapi.js in Action teaches you how to build modern Node-driven applications using hapi.js. Packed with examples, this book takes you from your first simple server through the skills you'll need to build a complete application.
الدورات التعليمية
-
Divide and Conquer, Sorting and Searching, and Randomized Algorithms
-
المشروعات
عرض ملف Matt الشخصي الكامل
-
مشاهدة الأشخاص المشتركين الذين تعرفهم
-
تقديم تعارف
-
تواصل مع Matt مباشرة
ملفات شخصية أخرى مشابهة
استكشاف مزيد من المنشورات
-
Mateusz Cedro
University of Antwerp • ١ ألف متابع
How a fun side project on AI Alignment became the cover of the Communications of the ACM, Association for Computing Machinery magazine! Every day we make plenty of small decisions that involve trading our comfort to save some money. Nevertheless, nobody really thinks that those decisions have some greater importance, as they look minor and we all have some internal preferences and a "decision boundary" where we can accept some level of discomfort to save some money. However, as we delegate more and more of our daily decisions to AI assistants, understanding how AI models value our inconvenience stops being a small question, as in the end the decision taken by an autonomous AI will affect our real comfort. Thus understanding and finding possible problems that may arise from these daily tasks are of real and great importance. Our article, "Cash or Comfort? How LLMs Value Your Inconvenience", investigates LLM behaviour in cash-or-comfort scenarios from the alignment and safety angle. At first glance the results look sensible and safe. Models clearly hold some abstract representation of comfort and money, and most of them produce a clean, monotonic decision boundary between cash-or-comfort trade-offs. On a closer look, however, we see plenty of misaligned behaviours in LLM responses that can cause real harm to a user. Among plenty of issues found, for instance, some frontier AI models will make the user wait an additional 5 hours to save €1. The same models, in other conditions, refuse a €1,000 reward for no inconvenience at all! Other models, for the same amount of inconvenience, will accept 9 euros of reward, then reject the same trade-off for 10 euros, and accept it back for 11 euros. Also, we found that adding a single whitespace character in a place where it carries no semantic meaning at all flips the model's decision from accept to reject. Interestingly, the chain-of-thought strategy partly solves unexpected behaviours, though it exhibits other issues. Also, scale makes a difference: the number of model parameters results in totally different assessments of the same trade-offs (check the article to see how!). Our results split frontier LLMs into cheap models and expensive models, and that difference is invisible until someone is already living with the decision taken. Those inconsistencies become dangerous the moment a user relies on the assistant blindly, because none of them are observable before the decisions are made. Our results call for far more scrutiny of daily LLM safety and alignment than it currently gets. A great collaboration with Timour Ichmoukhamedov, Sofie Goethals, Yifan He, James Hinns and David Martens at the University of Antwerp, with support from Flanders AI Research Program / Vlaams AI Onderzoeksprogramma and Research Foundation Flanders - FWO. Paper: https://lnkd.in/eAugRJBW #AIAlignment #AISafety #AgenticAI #LLM #CashOrComfort #CACM
٣١
١ تعليق واحد -
Ravit Jain
The Ravit Show • ١٧٢ ألف متابع
Declarative beats duct tape. I spoke to Eddie Jaoude, Lead Education Engineer at Kestra at the Big Data London, to unpack how teams actually ship with modern orchestration. Here is the flow we covered: • Why defining workflows declaratively changes the game for developers coming from Python DAGs • How teams adopt Kestra step by step without a risky forklift migration • Using the Playground in Kestra 1.0 to test steps safely and tighten the feedback loop • Plugging into existing stacks with plugins and a language-agnostic design so you do not rewrite everything • The first usage patterns Eddie sees in the wild, from scheduled jobs and data pipelines to event-driven tasks Clear takeaways. Fewer moving parts. Faster iteration. Stronger reliability. #data #ai #kestra #bigdataldn #theravitshow
٢٠٨
٢٢ تعليق -
Satya KONDAPALLI
ambariCloud • ٨ آلاف متابع
🥃 😡 Don't drink and drive; 🚗 Drive and Code: openclaw calling CoCo from WhatsAPP 🧠 In my openclaw, the 'ambarisnow' skill is a wrapper around Snowflake's Cortex Code CLI. ⚽ What it does: Sends a natural language prompt to the cortex CLI and returns the result. Examples: /ambarisnow show top 10 rows from AGENT_ORDER_LINES table Essentially the same as your coco alias but wrapped as an OpenClaw skill for use within that framework.
١٣
١ تعليق واحد -
Shail Shah
Electronic Arts (EA) • ٢ ألف متابع
One of the most interesting things about modern LLMs is that their biggest improvements aren’t obvious. Pre-Training large language models used to be fragile: Loss spikes, Instability, Sudden divergence. What changed wasn’t just better hardware only, it was architecture too. Pre-normalization. RMSNorm. Gated FFNs. RoPE. QK-Norm. Individually, these seem like small changes. Together, they’re the reason today’s LLMs can scale reliably. I wrote Part 1 of a deep-dive series breaking down these LLM architectural shifts and why they matter in practice. https://lnkd.in/eTkaxUZM Part 2 will cover attention computation optimizations and long-context scaling. #LLM #AIEngineering #MachineLearning #DeepLearning #GenerativeAI
١٦
١ تعليق واحد -
Danny Quinn
National Cloud UK • ٩ آلاف متابع
I've been using Claude's Projects, Skills and Claude Code for a while now and it's an entirely different experience to ChatGPT. Claude Code isn't just for coding, i an not a coder at all... What I actually use it for is building workflows, automating tasks and creating systems. Pair it with a note tool like Obsidian and you can genuinely create a team of AI agents working for you. Claude feels slightly more technical at first. That puts some people off. But once you get your head around a few basic concepts it's surprisingly simple. Zero coding knowledge required. My current weekly AI breakdown: → 60% Claude suite → 35% Gemini and tools like NotebookLM → 5% ChatGPT That ratio would have been unthinkable a year ago. There are hundreds of fantastic youtube tutorials on how to use Claude.
٥٠
١١ تعليق -
Maxime Gréau
Chainguard • ١ ألف متابع
🎉 Announcing DriftlessAF — Open Source Agentic Reconciliation Framework I'm super excited to be part of the team, alongside Matt Moore, Nghia Tran and Manfred Moser, that just open sourced DriftlessAF, the framework we use at Chainguard to build and run reconciler bots at massive scale. This is what powers Chainguard Factory v2: 2000+ container images, hundreds of thousands of package versions, CVE patch backports, all reconciled automatically DriftlessAF was built from the vision of Matt Moore, Chainguard CTO, who led a lot of the controller infrastructure for Knative. The idea: take that same reconciliation approach and apply it beyond Kubernetes: to GitHub repositories, OCI registries, APK packages, and more. ✨ What makes it different? DriftlessAF brings first-class support for agentic AI into the reconciliation loop. Traditional reconcilers are great for deterministic tasks. But some problems need human-like judgment: understanding code changes, writing meaningful descriptions, deciding how to fix issues. That's where AI executors come in. You can plug in Anthropic Claude or Google Gemini via Vertex AI and let the agent handle the fuzzy parts while the reconciler drives the loop. I recorded a quick demo based on the examples we published: - PR Validator: classic reconciler pattern (detect drift, report it) - PR Agent: reconciler + Claude Sonnet 4.5 (detect drift, auto-fix with AI) 📦 Examples: https://lnkd.in/eGvUkA4E 🔗 GitHub: https://lnkd.in/eVdqKJGz It's all Go, runs on Cloud Run, and we've been running it at scale for a while now. Excited to see what people build with it. #Reconciler #agenticAI #CloudRun #OpenSource
٦٦
٤ تعليق -
Kathleen DeRusso
Elastic • ١ ألف متابع
Chunking and snippet extraction has been a huge focus lately - my latest blog dives into some of the work we've done on this to date, including support for a chunk rescorer in our semantic reranking retriever, as well as some useful ES|QL primitives to get more visibility into chunks and snippets. #elasticsearch #semanticreranking #snippets #chunks #chunking #esql https://lnkd.in/e6UD7iii
١٣
-
Emeric Tabakhoff
Indépendant • ٥ آلاف متابع
Chris Travers: Introduction to NUMA Wrestling with large PostgreSQL systems can feel like taming a wild beast. Enter Chris Travers—our hero busting the NUMA myths. He starts his witty saga on Stormatics' blog. Apparently NUMA stands for Non-Uniform Memory Access. Who knew right? It's the funhouse mirror version of memory architecture making your processors and mainframes think hard. Once upon an early 1990s time our systems played nice with Uniform Memory Access until our fancy processors demanded more. Enter NUMA stage left favoring some memory over others just like my cat preferring one shoulder to the next. The result? Higher parallelism which means more chaos (read fun) in accessing memory but better performance too! NUMA means dividing system memory between memory controllers. The cores get a workout running across controllers—picture an interconnect relay race. And Linux plays the ringmaster here with policies named like flavors of ice cream: Local Membind Interleave. Chris hints at Postgres 17's adventures in NUMA land coming soon. If you survive this tech thriller, you're ready for that sequel. Keep an eye on Stormatics but remember—reading with humor might help. Props to Chris Travers! Check out his full shenanigans: [https://stormatics.tech]. You can find the full article here: https://postgr.es/p/7ka
١