Maz Shar

𝗦𝗲𝗻𝗶𝗼𝗿 𝗠𝗮𝗰𝗵𝗶𝗻𝗲 𝗟𝗲𝗮𝗿𝗻𝗶𝗻𝗴 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿 – 𝗗𝗶𝗳𝗳𝘂𝘀𝗶𝗼𝗻 𝗠𝗼𝗱𝗲𝗹𝘀 𝗕𝗿𝗶𝘀𝘁𝗼𝗹/𝗟𝗼𝗻𝗱𝗼𝗻 𝗨𝗽 𝘁𝗼 £𝟭𝟬𝟬𝗸 ▪ Revenue-generating AI start-up. ▪ Unique product with AI at its core. ▪ Recently raised multi-million ££ funding to support the next growth phase. Bristol is crying out for more AI companies so I’m buzzing to be supporting the latest player in this space. I’ve got two roles, one of which is more research based so I’m looking to speak with Machine Learning Engineers and Researchers with a solid understanding of diffusion models. The ML Engineer needs to have strong production experience and come from an image-based background. Diffusion model experience is the icing on the cake, but the client is aware there’s a limited number of people who will have experience taking these into production. Essentially, if the researcher has the required diffusion model expertise, then it’s less important for the engineer. There is a world whereby we can hire 2 x ML Engineers in place of a researcher, so long as the diffusion model experience is covered. 𝗘𝘅𝗽𝗲𝗿𝗶𝗲𝗻𝗰𝗲: ▪ 0-1 experience is preferred – failing that, experience within an AI startup and the ability to wear many hats. ▪ Strong computer vision background, ideally diffusion models or transformers at the very least. ▪ ML production, pipeline and infrastructure (design and build). ▪ Understanding of MLOps principles. The engineer position needs to be at the senior level, but the researcher is more flexible. As an ML Engineer, you’ll essentially be the subject matter expert, so you’ll need experience in designing and building ML systems and infrastructure from scratch. This could in time turn into a leadership position depending on your career aspirations but it’s not a deal breaker if you want to remain an IC. The preference is to build the team in Bristol but there is scope for those in London to apply as long as you’re happy with monthly travel to Bristol. If you’re based in Bristol, then it will be 3 x days a week in the office. Unfortunately, sponsorship is not available for either position. #AI #artificalintelligence #diffusionmodel #diffusionmodels #machinelearning #ML #mlengineer #aiengineer

16

1 Comment

We've published another great article to the John Lewis Partnership Software Engineering Blog. "5 lessons I learnt during an AI hackathon" was written by Mark Winteringham who is a Senior Quality Engineer in the Partnership. The article describes Mark's participation in a recent Partnership AI hackathon. In it Mark shares some reflections from a Quality Engineer's perspective on the two fantastic days of hacking. https://lnkd.in/eAfFFR6G #JohnLewis #AI #QualityEngineering

39

SEIS/ EIS cheatsheet for Founders (If you want to have a chat for guidance on this, DM me. We've recently successfully been approved for this scheme, and completed our Seed funding round) EIS qualifying criteria and limits - https://lnkd.in/eCNWny3z SEIS qualifying criteria and limits - https://lnkd.in/e82gefXX You can apply to qualify for both SEIS and EIS at the same time, so might as well go for both For your application and paperwork, I would recommend SeedLegals. They have a brilliant online portal you can use, as well as experts on hand to guide you throughout your process. Charges are reasonable too. For employees, you should look at EMI Share Options, which allows you to give employees a part of the company via options where you can set the rules, plus extremely tax efficient for them. For NEDs, you can do Unapproved Share Options, again you can set the rules but with less of a tax advantage. Typically you would give NEDs 0.5-1% to act as your NEDs in the first year. For both, Seedlegals can arrange all paperwork for you via their online portal. To raise, you can use sites like Crowdcube & Seedrs, but be aware they do charge a fee, normally as a % of the raise. There are also a ton of SEIS/ EIS VC firms that are always looking to invest, but they're a lot more involved in the day-to-day operation. VC's are great if you need that structure and guidance. Alternatively, you can raise from your network. Look for people earning £150k+, as the tax advantages for those individuals are very good indeed, and they tend to already be investing in such deals and have a portfolio of them, and hence understand the risk and timelines involved.

6

Reposting to help raise awareness of this unique opportunity to explore #OpenData in an #EdTech environment, with a side of #AI . If you are interested, I encourage you to read the blog post https://lnkd.in/eRjM3K7D

5

#GenAI is one of the #HotTopics within #softwareengineering right now, so if you'd like to hear the thoughts of Dom (CTO @ Cameo) and participate in a healthy discussion with like minded peers, just drop me a message!

4

Loving Jamie's article on the North East being a great place for business. Then, I would.... I travelled 325 miles to base Myfolks there. Of course, I wanted to help create wealth in and for the north east (as many know, our great Tyne Bridge is represented in the Myfolks' logo). But, I've been building commercial wraps around tech products for 25 years before I founded Myfolks and I wouldn't have made that emotional choice if it didn't make sound economic and commercial sense. The North East tech scene is navigatable - one of the first places I was pointed to is Sunderland Software City, from there to Paul Lancaster's events (check put Pl_tform) and on to the Innovation Supernetwork and onwards. The Universities, including my beloved Northumbria, Newcastle, Sunderland and Teeside are doing great work in the tech space, both in the north east and beyond and they work together - with industry, with founders and, crucially, with each other. Why wouldn't you want to benefit from the National Innovation Centre for Data in today's data driven world? Where would Myfolks be without the support of the National Innovation Centre for Ageing, their neighbours, located on the Helix in Newcastle and Health Innovation (North east and north Cumbria)? I hurtle about cramming in meetings in a way I couldn't in the south, often on foot, or the metro. You can do more in a day. The business community and the founders' community are so strong, helpful and collaborative. Plus, there's at least 7 Greggs in Newcastle and, did you note the tech Sela introduced recently to enable deaf kids to experience the noise and passion of the match? Makes you proud... Give Jamie's article in our fabulous QT a read.... See you in the North East soon...

13

2 Comments

Who’d have thought your GCSE in English would end up being a tech skill? 🎓📚 Every few years in the tech industry we find a new way to rebrand the word ‘engineer’, and LLM development has been no different in the rise of the ‘prompt engineer’. 🛠️ To be totally frank with you, when the phrase was first coined I scoffed at it. I’m from a world where model optimisation and tweaking is done through numbers, hyperparameters, and feature selection - how powerful could words actually be? 🧐 Slowly it’s dawned on me just how deep these deep learning models are, and how big the scale they’re running at truly is. 🌌 So what problems do we have? Security and safety are the two biggest ones - we don’t want our models to divulge potentially damaging secrets, and we also don’t want them to be wildly offensive to our users. This is a particular challenge given that these LLMs are trained on the internet (which can be a complete toilet), and also have no notion of tone, context, or offence. Early experiments into unleashing language models unfettered onto the internet had to be shut down extremely quickly as their responses became more and more unstable - ask Microsoft how their experience with Tay went! 😳🛑 What levers can we pull though? In more ‘traditional’ machine learning we can account for known bias using balanced sample sets, feature modification, and equalised odds prediction to level things out or disallow predictions we know to be problematic. However, LLMs are so large with such a huge training set that these models are not interpretable. 🧩 To solve the problem, therefore - the only thing we can do is modify the inputs - we have to ‘prompt engineer’. Which means that we have to craft our language in such a way that the most likely outputs are those which we intend. Because the models have been trained on something resembling the corpus of human knowledge, we have to write with intent, purpose, and tone - all the things your English teacher was getting at when you were 14. Models have to be literally told to not divulge sensitive information, to not deviate from their path, and not to be offensive or libellous. Prompts come with ** asterisks ** for emphasis, and STRONG LANGUAGE to indicate imperatives. ✍️ Even now the burgeoning profession of prompt engineers swap tricks and nuance to get LLMs to make better predictions (notoriously, pretending to offer Chat GPT a tip worked for a while). I never thought that I’d reopen my notes from my GCSE, but here we are. 📖 ❗️ Want more insights like this? Check out our free, Ultimate Guide to AI. Hit the link in the comments! 🔔 Want more of this content in your feed? Don’t forget to hit the bell on my profile.

17

9 Comments

I attended an eye-opening talk at #BirminghamTechWeeks #ScaleUpSummit this morning by Dr Abena Dadze-Arthur The key takeaway? While we're all chasing the perfect tech stack and processes, we're overlooking our biggest scaling lever: People's mindsets. A critical pattern that was noticed in successful scale-ups: 1. Growth mindset isn't just corporate jargon. Teams believing in continuous improvement consistently outperformed those stuck in their ways. Rather brilliant, when you think about it. 2. The "ownership mindset" spreads like wildfire. One team member starts treating the business as their own, and before you know it, the whole organisation is thinking about optimisation and long-term impact. 3. Most fascinating was the distinction between problem-spotters and problem-solvers. In rapidly scaling companies, the difference between "that's broken" and "here's how we fix this" proved decisive. Whilst competitors can replicate your systems, they simply cannot duplicate your team's collective mindset. #ScaleUp #Leadership #OrganisationalCulture #BusinessGrowth #TechWM

15

📚🔬 Revisiting My PhD Journey with NotebookLM! 🎓⚡ Just explored NotebookLM on my PhD thesis: "Essays on Risk and Profitability in the Future British Electricity Industry" 🇬🇧💡 Key takeaways: Amazed at how well it captures nuances 🧠✨ Brings back memories of the research grind (and breakthroughs!) 💪🔍 Fascinating to hear my work come alive 🎧🗣️ Want to experience it yourself? Check it out: 🔗 https://lnkd.in/eabdGCdf #PhDLife #ResearchTech #AI #EnergyEconomics #NotebookLM

8

I read an interesting article on the BBC News website this morning so decided to blog about it on my personal blog. I pasted the URL in and hit go! Seconds later I had a completely original version, so I edited it into my own voice and got sBlogIt! to create the images, audios and social media for it. And here it is ... https://lnkd.in/eD4a8TJa Want to be able to create blog posts in minutes? #sBlogIt!

11

2 Comments

Tomorrow I’ll be hosting the first of a series of webinars to explain why the Flexible AI Upskilling Fund is for you 🫵 If you are a professional, based in the UK working for an SME, you may be eligible for up to £10k in grant funding to train your people in AI. In this webinar I'll cover: • What AI could do for your business • What is UK government's 'Flexible AI Upskilling Fund' • Eligibility and funding details • The application process & how to apply • How to select the right AI training I'll also be releasing exclusive content, to help you make a successful application. If you don’t think AI is for you, think again. You don’t need to be able to BUILD AI systems to USE AI systems. AI is for everybody and I’ll show you why. Register for the webinar here: https://lnkd.in/egTtwDuT

33

2 Comments

Some really excellent talks this morning at BSides Leeds, especially this one on topics like privacy removal in the ioT world (many thanks Lianne P.!) Scary but important to think about! 

6

2 Comments

Shashi Verma hilighted some big truths I feel well worth sharing in a fantastic presentation on innovation at the Intelligent Transport Systems UK Annual General Meeting yesterday at Amazon HQ. Alongside some great nerdy facts about the history of ticketing at Transport for London (I feel well armed for the next pub quiz trivia round), he told us that: Firstly - it is possible, but very hard and very complex to make innovation happen at massive scale: "Leading this sort of change is lonely and needs enormous faith, courage, risk taking and resilience". Secondly - innovation in this space takes a catalyst: "Innovation in public sector only ever comes from 3 places: Defence, Change of regime and Mavericks" Shashi was (and still is!) a maverick, implementing contactless payment in London. We should be encouraging more mavericks to take big swings at the big challenges facing the UK now more than ever, because... ...Thirdly - "the only thing this country can think about is misery right now.... Everything has been made harder by the current fiscal situation. Organisations have to make things work with less money. That's the challenge". What I choose to take away is: there is huge opportunity in the mobility and infrastructure space for ambitious innovation, which can drive productivity. Implementation of big change in complex industry is incredibly challenging, and takes a large amount of grit, but there is precedent for long-lasting impact that the UK has made on the global stage (e.g. leading the charge of contactless payment). It has always been a privilege to participate in the conversation and it is encouraging to hear the clarity with which the CTO of (arguably) the best public sector transit organisation in the world is calling for action for infrastructure operators to embrace the challenge of innovation. #innovation #ticketing #aiforproductivity #problemfirst #datadriven Max Sugarman - thank you for hosting!

37

3 Comments

While researching Crypto Grant Programs 🌐, I discovered that the entire grant ecosystem suffers from the same problems. Whether you're searching, applying, or trying to secure funding for your scientific research, educational initiative, art or for-profit project, the process often becomes a frustrating burden. There's a significant lack of transparency and individual feedback, leading to uncertainty and no guarantees of success 😕. In the example below 46 funded projects out of 1645 reviewed applications makes it 2.79% success rate which in its turn makes 97.21% of wasted resources. I explored several ideas and did basic research on this topic. During my investigation, I found a small crypto grant program focused on writing 📝. Excited to apply, I noticed that one of the grant topics listed was the current crypto grant market landscape — a topic I had extensively researched. It seemed like the perfect opportunity to contribute to the market research 📊. I applied through their website, providing a detailed breakdown of my research plan and sharing links to prior in-depth research articles and videos I had created on various web3 topics. The grant guidelines promised a review and approval within 1-2 days (up to 48 hours). However, after three days with no response, I decided to send a reminder 📅. Eight days later, they finally responded with a rejection, explaining that I lacked the experience to deliver the research 🙄. Then, they removed the grant profile from their website completely 😲. Since starting my research in May 2024, I realized that grant programs can be a great way to drive innovation in the web3 ecosystem 🚀, especially within foundations. However, the current processes and standards pose a high risk of fraudulent and corrupt activities, affecting grant seekers, reviewers, and grantees alike ⚠️. Issues like time consumption, domain knowledge requirements, and the lack of feedback and transparent fund management can be solved with an AI-powered application running on a decentralized infrastructure, this architecture approach must be at the core of any modern grant management system 🤖🔗. #CryptoGrants #Web3Innovation #GrantTransparency #BlockchainResearch #DecentralizedAI #FundingOpportunities #AIinGrants #CryptoFunding #GrantManagement #AIforWeb3 #BlockchainGrants #InnovativeFunding #ResearchGrants #TransparentFunding

4

4 Comments

Absolutely enormous sponsorship opportunity at [possibly] one of London's largest in-person JavaScript meetups, London.js 🚨 We're looking for a sponsor for our June London.js meetup which is being hosted by the team over at Rightmove. We're still 2 weeks away but it's already at capacity with 180 people signed up and a waiting list that is only going in one direction 📈☝️🛫🆙 If you sponsor, it'll look a little something like this: 𝗟𝗼𝗻𝗱𝗼𝗻.𝗷𝘀 𝗵𝗼𝘀𝘁𝗲𝗱 𝗮𝘁 𝗥𝗶𝗴𝗵𝘁𝗺𝗼𝘃𝗲, 𝘀𝗽𝗼𝗻𝘀𝗼𝗿𝗲𝗱 𝗯𝘆 *𝘆𝗼𝘂𝗿 𝗰𝗼𝗺𝗽𝗮𝗻𝘆* - 𝗧𝗵𝘂𝗿𝘀𝗱𝗮𝘆 𝟮𝟬𝘁𝗵 𝗝𝘂𝗻𝗲! ...pretty cool, eyy? Typically, the companies that like to partner as a sponsor are those that are smaller in headcount with lack of space to host, remote first or earlier stage but looking to get some solid reach for their brand. But it really could suit any company using JavaScript in one way or another. So, what do you get in return? 💛 Recognition as a company that involves itself in the tech community in which you operate. Big tick ✅ 🤝 Allocated a slot before the evening talks start to introduce your company & the unbelievable work you're doing 📣 Given one of the technical talk slots for someone in your team to talk about a topic of their choice, linked to product or engineering 🔌 Alongside all of this, if you're hiring, it's a great time to drop-in that you hire Software Engineers to a room full of Software Engineers 👀 Some of our previous sponsors include: Birdie, Netflix, Twilio, NewDay, PwC, loveholidays, incident.io, Blink - The Frontline Experts. If you're interested in hearing more, just show me a sign in a method of your choosing. Here on LinkedIn is preferred but I can be flexible. 🙏 James McLeod Jordan Potts Ana Sampaio Tay Bencardino Liam Abkal-Thomas #londonjs #javascript #meetup #community #london #sponsor #sponsorship

29

1 Comment

"The Power of Tomorrow" is something I often think about. Most of the technologies we have now will only improve, in scale, speed and function. Of course, our usage of those technologies could accelerate a technological higher human purpose, or just lead to terminal enshittification, model collapse, overall platform decay. In this article for Silicon UK I discuss, along with several industry voices, some of the technologies that will revolutionise our lives in the next decade. Thanks to Andy Shaw for getting me into this publication. https://lnkd.in/e2REciKD #quantumcomputing #edgecomputing #ai #artificialintelligence #cybersecurity #futurism

14

1 Comment