Lon Pilot (he/him)

From: CEO, To: CPTO Subject: Headcount Freeze... Your mind races as you consider how you are going to hit targets. A sense of panic sets in. What about my open VP Engineering (VPE) vacancy? That one can't be frozen. Surely not... You call your CEO. → 𝗖𝗣𝗧𝗢: Hi Charlotte, Just got your message. I intend to progress with the VPE hire as agreed last week. → 𝗖𝗘𝗢: I'm glad you called. I wanted to discuss that role in person. The VPE is included in the freeze. -- 😱 You are speechless You are from a product background A small "t" CPTO You 𝗻𝗲𝗲𝗱 a VPE to help mature engineering and ensure technology is having as much 𝗯𝘂𝘀𝗶𝗻𝗲𝘀𝘀 𝗶𝗺𝗽𝗮𝗰𝘁 as possible. You have enough challenges to deal with in product. This was acknowledged and agreed with the board You are not sure how much longer you can cope without a VPE -- → 𝗖𝗘𝗢: Can the Engineering Managers help? → 𝗖𝗣𝗧𝗢: I don't think so. We did not expand this layer as our team grew. We should have but we didn't get budget approval. This has left Engineering Managers with large remits and workloads are becoming unmanageable. Furthermore, Tech Leads are now expected to line manage, many of them doing this for the first time. Engineering Managers are spending the vast majority of time supporting first time managers. They are operational I need strategic. Hence the VPE. → 𝗖𝗘𝗢: Do you have any other suggestions? → 𝗖𝗣𝗧𝗢: I'll have a think and get back to you. -- 🌟 Hi CPtO 👋 I'm Ryan, The Product Minded CTO and I am going to take your pain away. Budgets are tight. I get it. What if I told you that access to 𝗽𝗿𝗼𝘃𝗲𝗻, 𝗶𝗺𝗽𝗮𝗰𝘁𝗳𝘂𝗹, 𝘀𝘁𝗿𝗮𝘁𝗲𝗴𝗶𝗰 engineering 𝗹𝗲𝗮𝗱𝗲𝗿𝘀𝗵𝗶𝗽 can be achieved for a 𝗳𝗿𝗮𝗰𝘁𝗶𝗼𝗻 of the 𝗰𝗼𝘀𝘁 of your proposed perm VPE? 𝗜𝘁'𝘀 𝗰𝗮𝗹𝗹𝗲𝗱 𝗙𝗿𝗮𝗰𝘁𝗶𝗼𝗻𝗮𝗹 𝗟𝗲𝗮𝗱𝗲𝗿𝘀𝗵𝗶𝗽 → 𝗖𝗣𝗧𝗢 How does it work? I'll come into your organisation several times a week and lead engineering for you. • I'll seek to understand your business goals and assess current state of engineering • Work with the engineering managers to create an engineering strategy that unlocks your desired outcomes • Train your Tech Leads how to be great people managers and team leaders 𝗜𝗺𝗮𝗴𝗶𝗻𝗲 𝗮 𝘄𝗼𝗿𝗹𝗱 𝘄𝗵𝗲𝗿𝗲… → 𝗣𝗲𝗼𝗽𝗹𝗲 are inspired, aligned and performing to a high standard → 𝗣𝗿𝗼𝗰𝗲𝘀𝘀 right sized enabling you to work effectively at scale → 𝗧𝗲𝗰𝗵𝗻𝗼𝗹𝗼𝗴𝘆 an asset rather than a liability. 𝗥𝗲𝘀𝘂𝗹𝘁𝗶𝗻𝗴 𝗶𝗻... • Innovation driving top line growth • Costs actively managed • Risk within appetite. 𝗕𝗲𝗻𝗲𝗳𝗶𝘁𝘀 • Flexible (I'll stay as long as you are getting value) • Affordable (No bonus, equity, benefits) • No long term headcount commitment • Outcome focussed • Can start tomorrow! 𝗗𝗠 "𝗛𝗘𝗟𝗣" to continue the conversation

9

I see a lot of posts on LLMs that decry their usability. They mainly focus on off-the-shelf solutions, like ChatGPT and Gemini. Typically, they offer up examples of chat applications failing in one way or another to accurately answer industry-specific problems, and they point to these errors as evidence of ineffectiveness. In my opinion, they miss the point. Chat applications are the broadest possible deployments of this technology. The real power comes from using techniques like Retrieval-Augmented Generation (RAG) to enhance the accuracy and reliability of LLM models with facts fetched from external sources. Using frameworks like Langchain and the Llama Index, and platforms like GCP, you can build truly transformative applications, leveraging your own specific knowledge base. You only have to look at tools like GitHub Copilot to see the effective use of LLMs and RAG, which can deliver 40-50% increases in productivity. In short, consider LLMs as an underlying technology that can be leveraged and improved with frameworks and techniques, and the possibilities for its use only grow! #llm #chatgcp #openai #gemini #gcp #rag

9

That pesky ECDSA signature refuses to go away, and we are struggling with it in a distributed world of trust. Luckily EdDSA is much more scalable, and now used in many distributed signature applications (and where we can split the private key up into secret shares and distribute them over several nodes). But we have a problem ... elliptic curve methods are going to be cracked by quantum computers. So can we take our PQC signature methods and scale them to distributed signatures? Well, no, as Dilithium doesn't support it. But Raccoon can, and also has better side channel resistance (and with a similar performance and key size to Dilithium. So, meet Raccoon: https://lnkd.in/d2QMHSM9

30

2 Kommentare

TRUST TAKES YEARS TO BUILD SECONDS TO BREAK, & FOREVER TO REPAIR One of the biggest risks posed by Gen AI is a further & irreversible erosion of trust. How can technology help counter this risk?

22

2 Kommentare

Great overview of 3 powerful LLM development frameworks. I think langchain is the most mature but I do like the embedded sub-chat capabilities of autogen.

4

Machine Learning Engineer - Founding Team - London £200k If you answer yes to 4/5 of these questions, then you want to check out this job (link in comments) 1️⃣ Do you have experience with quantization techniques (e.g., PEFT, DeepSpeed, TensorRT) to optimize ML models? 2️⃣  Have you ever managed or worked with a multi-node HPC cluster? 3️⃣ Are you familiar with Deep Learning using PyTorch Library? 4️⃣ Do you have experience with Kubernetes container orchestration? 5️⃣  Have you worked with some of the recent Large Language Models (e.g., OpenAI, LLaMA)? #LLM #ArtificialIntelligence #Quantization

21

3 Kommentare

Top tip for "No Code" integration solutions: Only use them if it can be done internally (i.e. someone in the business is has the technical ability to build and manage them) and the monthly platform bill is < £500. If you are paying a third party to build no code integrations - you may as well get them custom built so that you avoid the high ongoing fees and scalability issues that come with a no code platform. As the business scales you'll reach a point where it is more cost effective to do this anyway.

4

How to get the best out of your Senior Engineering Managers Be clear on performance expectations. Tell them how their performance will be judged. For Senior Engineering Managers, managers of managers, I favour a balance scorecard across people, business impact and technical excellence. It all starts with 𝘁𝗿𝘂𝘀𝘁. → Run your teams as you see fit with an ownership mindset → Operating constraints → Strategic context Now lets move on to the measures: 𝗣𝗲𝗼𝗽𝗹𝗲 Engaged as measured by • Employee satisfaction surveys    • Team health surveys    • 360 feedback    • Skip level 1:1s 𝗕𝘂𝘀𝗶𝗻𝗲𝘀𝘀 𝗶𝗺𝗽𝗮𝗰𝘁 Are teams: • Driving revenue growth    • Controlling costs 𝗧𝗲𝗰𝗵𝗻𝗶𝗰𝗮𝗹 𝗘𝘅𝗰𝗲𝗹𝗹𝗲𝗻𝗰𝗲 • Systems are stage appropriate and conform to the required levels of stability, scalability and performance.    • Technical risk/debt is actively managed and within appetite    • Close partnering with product. Technical innovation helps inform product direction. However you decide to judge performance remember... 𝗧𝗲𝗹𝗹 𝗺𝗲 𝗵𝗼𝘄 𝘆𝗼𝘂 𝘄𝗶𝗹𝗹 𝗺𝗲𝗮𝘀𝘂𝗿𝗲 𝗺𝗲, 𝗮𝗻𝗱 𝗜 𝘄𝗶𝗹𝗹 𝘁𝗲𝗹𝗹 𝘆𝗼𝘂 𝗵𝗼𝘄 𝗜 𝘄𝗶𝗹𝗹 𝗯𝗲𝗵𝗮𝘃𝗲 -- 𝘌𝘭𝘪 𝘎𝘰𝘭𝘥𝘳𝘢𝘵𝘵 P.S beware of unintended consequences of the measures you choose.  How do you measure the success of your Senior Engineering Managers?

4

Google's latest #AI search needs to go offline, and rebuilt, ground up. Again. AI is hard in product. It takes diligence and a focus on accuracy, facts, and integrity of how you trained it. Sadly Google has given two prime examples of what not to do - Gemini the factually wrong racist AI, and now search AI has corrupted results. You can find more examples ... this is the least scary, as I don't want to repost the answer to depression it gave users. 👉 My teams spend huge amounts of time testing our data. We have vigorous cyber efforts, audits of data origination, privacy and legal involved at design and release, and massive testing before BETA. We have had 2 successful product launches that have saved real productivity to our customers. 100s are in the works and massive transformation is occurring internally as we expand our capacity. All to say - you CAN make great products that work, serve your customers, and are fun for engineers to build. It requires the right culture; strong work ethic, and intense cross partnerships. Please build great AI. We will all have blunders, best we learn and grow than sink the boats in frustration.

1

https://lnkd.in/eq8zgEsE Very nice article. We don't spend too much time looking into academic fraud. However, data coming from high brand institutions and profiles carry more weight on the AI models and we may need to start paying attention to what's being said by top professors and institutes in today's world of AI. #acfe

16

I frequently emphasize the need to study Domain-Driven Design deeply. If you have a business challenge, it's essential to invest in obtaining a suitable and customized solution. The domain is the heart of every business and often brings out the details of each context. We always use generic themes in presentations for the same reason. A good source for reading are the books by Vaughn Vernon and Eric Evans. 🏃🏆🏆🏆🏆 #software #ddd #domain #domaindrivendesign

2

1 Kommentar

This is a very interesting article on sparse vectors, which can help overcome limitations of vector search related to domain specific keywords. And of course, the code in the example is clear and clean! Sparse vectors have a few values and many zeros. Dense vectors have limited or no zeroes. But really, what we're talking about semantic search on terms where the context is not present or harder to infer. The author presents this as specialized keyword search such as medical terms or other domain specific terminology. "These scores reflect the significance of each term in the context of the document or query, guiding the model to allocate more weight to terms that are likely to be more meaningful for retrieval purposes." Highly recommend reading. Nice work Paul Iusztin. And a nod to Qdrant and their vector search technology that supports sparse vectors. https://lnkd.in/gcBWfTg8

3

The new episode of Spilling The Source is out. this time I talk to Marco Vinciguerra about Scrapegraph AI. A site scraper that uses LLM models for processing. https://lnkd.in/eJ2m6APz

5

2 Kommentare

Sam Altman takes nuclear energy company Oklo public to help further his AI ambitions https://buff.ly/4bztCN4 - 💼 Sam Altman is the CEO of OpenAI and chairman of nuclear power company Oklo - 📈 Oklo went public through AltC Acquisition Corp., trading on NYSE as "OKLO" - 💰 Oklo received over $306 million in gross proceeds from the transaction - 💡 Oklo's business model is based on commercializing nuclear fission for energy needs - 🔬 Oklo is working on building its first small-scale reactor in Idaho - 🌎 Altman sees nuclear energy as crucial for meeting AI energy demands - 🔥 Altman has invested in other nuclear energy and AI-related ventures - 🤔 Altman believes nuclear energy is the best way to meet AI energy needs without fossil fuels - 🚀 Oklo CEO targeting 2027 for first plant to come online - 📉 SPAC market has seen setbacks but AI-related companies are gaining interest on Wall Street

🔄 Prioritize GenAI Projects Based on Your Maturity in Key Areas 🔃 I just finished giving this webinar jointly with Bern Elliot, who came up with the #GenAI "immaturity model" - like a baby, GenAI's age in the enterprise is measured in months, not years. The webinar is available for replay https://lnkd.in/gYXRATFt.

42

2 Kommentare

If LinkedIn post trending I've seen is to be believed: TDD seems to be making a HUGE comeback. It reminds me of when JS came back into fashion (a long time ago, kids) after being relegated to the pile of joke languages (an even bigger joke back then, JS-haters). For the longest time, it was "TDD solves everything and any opinions to the contrary are wrong". Then you heard nothing about TDD, not a peep. I think it was right around the time of DHH's post about TDD (below). Now, tada, it's back, and just like before, you're wrong if you don't have `I ❤️ TDD 4-EVA` tattooed on your lower back. Me and my team break all the rules and it would be hard (but not impossible) to convince me otherwise: - We only test end-user facing UX, full-stack/integration level. I'm not maintaining a test for MyServiceObject#convert_this if the method signature changes. I only care about the user experience. - Each daily PR must include coverage but big features behind a flag get coverage near the end. Each team member can do what they want but I'm not burning a day of complex mocking just so Product can say "whoops, that's not needed now". Coverage when it's showtime before the curtains go up. We are currently on Playwright + axe-core (for a11y compliance) and we love it. Migrating away from rspec + capybara (flakier than a bowl of frosted flakes). More on Playwright in another post (it's amazing). I don't think TDD is bad. I just want to mock things as few times as possible before a release, because test env data bootstrapping is a giant pain. To each their own, as long as your UX is protected.

9

4 Kommentare

Anthropic just launched Claude 3.5 Sonnet, which beats both OpenAI GPT-4o and Anthropic’s previous paid model Claude 3 Opus on most benchmarks. Claude 3.5 Sonnet can execute and debug code, and has improved image understanding capabilities, including text transcription. A new feature called “Artifacts” enables users to view generated code, documents and websites in a separate window next to the chat. Claude 3.5 Sonnet is available for free to all users, but paid users get higher usage caps. Did Anthropic just take the lead back from OpenAI? It seems that ChatGPT’s Code Interpreter tool, which used to be one its main advantages, has finally met its match, but ChatGPT still has the advantages of real-time Internet access and custom GPTs which Claude lacks… What do you think? https://lnkd.in/d_hk86cf

10

3 Kommentare

TDD haters are everywhere. For more than a decade, I have dealt with them, and I still do. The curious thing is that I could classify TDD haters into two categories: Group 1 - Practiced TDD for several months and several projects and they still don’t see the value. Group 2 - Tried TDD for two weeks and claimed it is garbage, unnecessary, and anti-agile. Here is another curiosity, based on my own empirical observations: Group 2 represents 99% of all the TDD haters I have to deal with.

8

2 Kommentare

On Michael Feathers' Keynote: Design Discovery in Existing Systems In his keynote at Domain-Driven Design Europe, Michael provided great insights into aligning code with its conceptual model. He stressed the importance of both top-down and bottom-up approaches. Michael emphasized that domain interrogation, a top-down method, helps gain better domain knowledge. Equally important is the bottom-up approach—learning directly from the code. This means understanding behaviors and concepts that might be scattered across the code. A key question he posed was, “Can we come up with a name for it?” A memorable point was his caution about metaphors, which can hinder understanding—a sentiment echoed by Eric Evans in his masterclass. Michael encouraged using surprises in the code as learning opportunities and mentioned Chesterton's Fence: “Don't tear the fence down until you know why it is there.” In my experience, e.g., modeling contractual models for German pharmacies required both approaches. Top-down modeling provided an overview, while the bottom-up approach revealed intricate details and special cases. This iterative swing ensured a comprehensive model. This keynote reminded me that effective modeling combines both techniques, allowing for incremental evolution. Trust in your ability to balance these approaches is key to successful system design. #DDDEU #SystemDesign #Modelling #LearningFromCode

7