Vlad Fomenko

A little late in posting the PRD+MRD. As promised but here it is in the form of a presentation. My document is lost somewhere and not able to find it but had captured it in the presentation. It is about an established company trying to launch and position itself as a gaming phone. Have you semrush and miro board as tools for personas and user journey respectively. Pls leave a comment or like incase of any suggestions you may have😀. #productmanagement #productmindset #productrequirements #gamingphone #miroboard #semrush

12

You don’t normally see LLM researchers and engineers talking about their day-to-day work. Big shoutout to this YouTube video from the Anthropic LLM Interpretability team, the team behind the Golden Gate Claude. It’s amazing to see that most of the team members have worked for the team for only one year, and yet they have made such a big breakthrough in interpretability research. Another thing I want to highlight is the unique challenges facing the team and how they approach these challenges, that is, through fast trial-and-error iterations and constant evaluation of what may work and what to prioritize next. #artificialintelligence #machinelearning #deeplearning https://lnkd.in/eUaPAYch

97

1 Comment

I'm a computer scientist who now works in biotech and writes poetic blogposts in Russian. How come? What do AI, poetry, and biology have in common? Language. At school I hated Russian grammar class. Learning a language by its rules is like learning to dance from a book. Even if you master a thousand rules and a million exceptions, it doesn't guarantee you'll write well. A natural language isn't designed; it's evolved. It wasn't developed, it developed. Studying grammar helps, but the only real way to master a language is through developing intuition – by reading, writing, and speaking a lot. Turns our the same applies to computers. For years, we tried to teach them natural language through complex logic, grammar rules, and it went even worse than my Russian lessons. Remember those dumb chatbots from a decade ago? But then, large language models (AI) arrived, and instead of programming on linguistic rules, we just had them "read" a lot, basically every text in existence. And now, ChatGPT writes in almost every natural language better than humans. But the most exciting part lies ahead. See, biology is also a language in a way — the Language of Life. Amino acids are the "letters," proteins the "words," biochemical pathways the "sentences," and every cell is an "epic" of 94 million words. This language, too, is the product of evolution, yet far more complex than Russian. For centuries, we've tried unsuccessfully to decode it by rules, and we've only managed to create something akin to the first chatbots. But soon, AI will change everything. At OPTIC we're building a translator at the crossroads of AI and biology. I believe that once we find an optimal, scalable way to collect daya and let machines read all the existing "texts in the language of life," we'll develop a model capable of writing in it. Can you imagine the possibilities when the "ChatGPT moment" happens in biology? A cure for every disease? Personalized therapy for everyone? An end to aging? — "Ok, BIOPTIC, cure Parkinson's disease." — "HDEKKPKNFFGWKDHEQCMHPKSIM…🫶" This is what drives me to work every morning.

441

15 Comments

Interesting thread on Twitter from Derya Unutmaz and Nick Dobos who think that medicine is going to crumble much faster to AI than any other industry. And they have good points: 1. Huge amount of gatekeeping and refusal to empower patients to make their own decisions. 2. Not an optional thing in many cases, even when you can’t afford it. So you are forced to use the next best thing. 3. Overpaid admin nightmare to get simple things diagnosed suddenly vanishes. What % of patients is that? -- "This is the final warning for those considering careers as physicians: AI is becoming so advanced that the demand for human doctors will significantly decrease, especially in roles involving standard diagnostics and routine treatments, which will be increasingly replaced by AI. This is underscored by the massive performance leap of OpenAI’s o-1 model, also known as the “Strawberry” model, which was released as a preview yesterday. The model performs exceptionally well on a specialized medical dataset (AgentClinic-MedQA), greatly outperforming GPT-4o. The rapid advancements in AI’s ability to process complex medical information, deliver accurate diagnoses, provide medical advice, and recommend treatments will only accelerate. Medical tasks like diagnosing illnesses, interpreting medical imaging, and formulating treatment plans will soon be handled by AI systems with greater speed and consistency than human practitioners. As the healthcare landscape evolves in the coming years, the number of doctors needed will drastically shrink, with more reliance on AI-assisted healthcare systems. While human empathy, critical thinking, and decision-making will still play an important role in certain areas of medicine, even these may eventually be supplanted by future iterations of models like o-1. Consequently, medicine is becoming a less appealing career path for the next generation of doctors—unless they specialize in intervention-focused areas (such as surgery, emergency medicine, and other interventional specialties), though these, too, may eventually be overtaken by robotic systems…maybe within a decade or so." Link to the thread in comments #healthcare #ai #innovation #technology #diagnostics #medical

1

2 Comments

Alex Komoroske's writing is so fun to read and so easy to draw lines into our reality: "Sarumans can’t see the value of systems; it will seem too fuzzy and abstract; [...] For similar reasons, a Saruman is deeply skeptical of other people’s perspectives on what’s a constraint, because the only constraints that are real are physical constraints, and everything else is just a fiction in people’s minds that can be overcome with boldness. Sarumans are the kinds of people who build impressive, powerful cathedrals. When you’re building a cathedral, you can’t have naysayers pointing out imagined constraints. You need decisiveness. As they get more powerful, accumulating followers, capital, and respect, it becomes easier and easier to cast the naysayers away. [...] Radagasts intuitively understand the power of the swarm: the system, in its buzzing, blooming, emergent glory. Radagasts know how to let go and dance madly with the system. They don’t try to control the system and its components. They love the system and its components: a love borne out of respect." https://lnkd.in/e9TNgNXb

2

Aysenur Ezgi Eygi, an American, was executed by an Israeli sniper in the West Bank this past Friday. She was shot in the head right after a weekly multi-religious prayer. Earlier this year, Tawfic Abdel Jabbar and Mohammad Khdour were killed in the West Bank by Israeli gunman. Both were 17 years old Americans. Journalist Shireen Abu Akleh, an American journalist, was executed in 2021 by an Israeli sniper while wearing a press vest in the West Bank. In 2003, Rachel Corrie, an American, was crushed by an armored Israeli bulldozer in the West Bank. This was ruled an "accident" by Israeli officials. As a reminder: 1. These are Americans being killed by a foreign nation 2. These Americans are being killed on lands deemed illegal by the US government 3. These Americans were not violent or posing any threat 4. The US has either not launched an investigation (ie the 17 year olds) or the investigation has gone silent 5. Israel has a "shoot-to-main" policy which a 2023 report showed that Israeli forces killed 10 Palestinians and over 6,000 protestors from May 2021-2022 To summarize: iInnocent Americans killed by a foreign nation on illegal lands = self defense + more military supplies. The genocide needs to end. https://lnkd.in/e-_Wu-Ee

1

Google had nine of my fellow Googlers *arrested* last night. Their "crime"? Peacefully protesting Google's Project Nimbus contract with the Israeli government, and recent revelations that, despite repeated explicit claims by Google to the contrary, this contract includes providing discounted cloud computing and consulting services to the Israeli Defense Force (IDF). You can agree or not with the overall message and goals of #NoTechForApartheid. You can debate whether it's appropriate to call what's happening in #Gaza a #genocide. But you can't deny that this was an act of pure #retaliation against these Googlers. Google claims that the removal of these Googlers was a matter of "workplace safety". Whose safety, exactly, was being compromised by nine peaceful employees inside the workspace of their own employer? You can say they should have used official channels - they did! We have! Hundreds of Googlers have spoken out about Project Nimbus internally through all manner of official channels. We have been consistently ignored, lied to, and given nonsense non-responses. Google culture explicitly encourages escalation - what is the next step when you've been ignored through all official channels but to make yourselves impossible to ignore? We do not consent to the output of our labor being used, directly and indirectly, to fight an unjust war, kill innocent civilians, and commit war crimes. We will not be silenced through illegal retaliation for speaking up about the conditions of our workplace. We will not be used. What can one person do? Use your voice. Speak up. And join with others. In collective action, we have power. Join the Alphabet Workers Union (AWU). Sign the petitions to end Project Nimbus, restore Eddie Hatfield's job, and more (internally or the external petition on notechforapartheid.com ). If you live in the US, contact your representatives in government and tell them that you want to stop funding this unjust war, and that you want clear violations of labor law such as Google's punished, and worker protections enforced and improved. Alone, it can seem like it's impossible. But together, we can make change. #Solidarity #Labor #Union #AWU

215

10 Comments

MarkovML was founded on the belief that your data is the ultimate competitive advantage. As Peter Norvig famously stated, "𝙈𝙤𝙧𝙚 𝙙𝙖𝙩𝙖 𝙗𝙚𝙖𝙩𝙨 𝙘𝙡𝙚𝙫𝙚𝙧 𝙖𝙡𝙜𝙤𝙧𝙞𝙩𝙝𝙢𝙨 𝙗𝙪𝙩 𝙗𝙚𝙩𝙩𝙚𝙧 𝙙𝙖𝙩𝙖 𝙗𝙚𝙖𝙩𝙨 𝙢𝙤𝙧𝙚 𝙙𝙖𝙩𝙖." This adage is particularly relevant in today's landscape, where Large Language Models (LLMs) are becoming commoditized. This has again been proven by the LLama3.1 paper (links in the comments). The difference between success and failure often hinges on having right data and how effectively it's managed and utilized by business. Here are some of the best practices to maximize value for your enteprise. 👉 𝗙𝗼𝗰𝘂𝘀 𝗼𝗻 𝗗𝗮𝘁𝗮 𝗤𝘂𝗮𝗹𝗶𝘁𝘆 ✨ Regular Audits: Ensure data accuracy and integrity through consistent and thorough audits. ✨Validation and Updates: Ensure quality metrics are in place to ensure consistency and relevancy. 👉 𝗨𝗻𝗱𝗲𝗿𝘀𝘁𝗮𝗻𝗱 𝗬𝗼𝘂𝗿 𝗗𝗮𝘁𝗮 𝗗𝗶𝘃𝗲𝗿𝘀𝗶𝘁𝘆 ✨ Inclusive Data Collection: Diversify your data inputs to ensure your datasets genuinely represent global perspectives and scenarios, enhancing the applicability and relevance of your models. ✨ Broad Representation: Strive for datasets that encapsulate a variety of demographics, industries, and use cases. This diversity helps improve the generalization of your models across different contexts and applications. 👉 𝗘𝗻𝘀𝘂𝗿𝗲 𝗗𝗮𝘁𝗮 𝗶𝘀 𝗨𝗻𝗯𝗶𝗮𝘀𝗲𝗱 ✨Bias Detection: Models are biased because the data used to train them is. Identify & fix the root cause. ✨Fairness Metrics: Establish fairness metrics and ensure your models deliver equitable outcomes across all user groups. 👉 𝗔𝗱𝗼𝗽𝘁 𝗮 𝗣𝘂𝗿𝗽𝗼𝘀𝗲-𝗗𝗿𝗶𝘃𝗲𝗻 𝗔𝗽𝗽𝗿𝗼𝗮𝗰𝗵 𝘁𝗼 𝗠𝗼𝗱𝗲𝗹𝗶𝗻𝗴 ✨Clear Objectives: Ensure you establish your expectations from the model and understand the trade-offs. ✨Alignment with Business Goals: Models scale efforts constrained by time and resources. Have clear business metrics and measure model performance against these to maximize value. 👉 𝗘𝗾𝘂𝗶𝗽 𝗬𝗼𝘂𝗿 𝗧𝗲𝗮𝗺𝘀 𝘄𝗶𝘁𝗵 𝘁𝗵𝗲 𝗥𝗶𝗴𝗵𝘁 𝗧𝗼𝗼𝗹𝘀 Give your knowledge workers the right tools to automate routine data tasks and efficiently transform raw data into strategic insights so the business can make informed decisions faster. Superior data and the right tools aren't just assets; they're your competitive edge—invest in them for long-term success. #DataQuality #DataDiversity #AI #MachineLearning #BusinessStrategy #MarkovML #GenerativeAI #GenAI #Llama3.1

31

3 Comments

Cost-aware inference is important, but currently LLM prompts are not interoperable. Prompts that work good on one LLM may show worse performance on the other, even more advanced LLM or even on the different versions of the same LLM. Once interoperability problem is solved it will also be possible to implement cost-aware inference that strategically manages computational resources for prompt processing, ensuring sustainable costs. This approach includes model routing https://www.notdiamond.ai/ - automatically selecting the most appropriate model for each task. Startups in this area are actively supported by the industry veterans.

4

2 Comments

Hugging Face demonstrated test-time scaling with Llama 1B, turning it into a "reasoner" that outperformed Llama 8B (8x larger!) in math. Test-time scaling has opened up a whole new dimension in model performance and AI research. I wrote about this when OpenAI released its o-1 model (the first 'reasoner' model). You can read it here: https://lnkd.in/gi737nsA At its core, a reasoner model uses Chain of Thought (CoT) reasoning—breaking problems into logical steps, solving them sequentially, and backtracking when needed. Scaling test-time compute enables this reasoning through two techniques: 1/ Self-Refinement: Models iteratively improve outputs, correcting errors over multiple passes. 2/ Search & Verifiers: Generate multiple outputs and evaluate them with verifier models (e.g., Process Reward Models for steps, Outcome Reward Models for final answers). Hugging Face’s contribution stands out because they’ve shown test-time scaling can be applied to any model, even smaller, open-source ones. They’ve also made it easier for others to replicate and advance the technique—helping the OS community catch up. That said, inference scaling is still far from efficient. Llama 1B required ~20 reasoning steps to outperform Llama 8B, a model only 8x larger. Nevertheless, inference scaling already achieves two critical things: – Pushing the frontier: Scaling inference pushes the boundaries of accuracy and delivers better results. – Smaller hardware requirement: Smaller models can match the performance of larger ones, reducing reliance on expensive infrastructure. The next frontier isn’t just about bigger models—it's about smarter inference. Excited to see where this leads!

62

4 Comments

I launched a dedicated API for web page summarization a year ago. At the time, I noticed that people were struggling to use ChatGPT (v3.5) to summarize URLs effectively. Nobody really understood how LLMs operate back then. There were ZDNet articles with instructions like "just put an URL into ChatGPT and it will output you the article summary" - the journalist didn't realize that ChatGPT hallucinated and generated summaries based on keywords found IN the URL - rather than the actual article text. To address this issue, I created a simple API that receives a URL to scrape as input and returns an article summary as output. ScrapeNinja handled all the heavylifting for web scraping part, utilizing rotating proxies, smart retries, and real browser rendering when required by the website. The LLM integration was straightforward – I simply passed the markdown of the scraped web page into GPT 3.5T, iteratively, making sure I fit into 16K token window. The API was an instant success, attracting thousands of customers overnight, from a single random YouTube tutorial on "How to launch an AI business." (the author just happened to use my API). I spent zero dollars on the marketing. Hundreds of customers are still using the API. It was an exciting time, indeed! I have just updated the underlying LLM of the API to Claude, to improve quality, reduce latency, and cut my costs.

30

8 Comments

Neural Machine Translation (NMT) vs. Large Language Model (LLM): Does NMT Still Have a Place in the Time of LLM? I’m not going to pretend to be an expert in computational linguistics or artificial intelligence, but I’m lucky enough to use both machine translation (NMT) and AI (LLM) every day. Plus, I get to see expense reports for a couple of companies (PIARA Inc. and New Paradigm LLC) every month. So, here are my thoughts: All three LLMs I use most often—ChatGPT (any version), Gemini, and Llama3—translate meaning better than Systran (v9), Azure Cognitive Services (Bing Translator), or Google Translate when dealing with "normal" text. By normal, I mean text that an average native speaker of the language can understand. The issue starts when the original text is "polluted" with niche slang. Hallucinations 🤥 make it a fun read, but they make the translation unusable. As a workaround, you can fit the vocabulary into the prompt, which helps, but depending on the number of possible source languages, it might not be practical. NMT translations don’t sound as natural as those from LLMs, which is kind of their strong side when dealing with niche slang. NMT retains enough “original” phrases for someone with expertise in the particular niche to understand what the original text talks about.👍 Another significant difference between NMT and LLM is the cost. To illustrate, I'll provide Google Translate pricing since they clearly state it for both NMT and LLM. 💰 NMT (Text translations) - $20 per million characters*  💰 LLM (Adaptive translation) - $25 per million characters* (input characters) + $25 per million characters* (output characters) LLM text translation is currently in preview and free, which makes it unsuitable for this comparison. https://lnkd.in/erdzbSxn So, as we can see, it’s $20 for NMT vs. $50 for LLM (if I understood the pricing table correctly). Depending on how much you use it, the difference could be significant or negligible. It gets more interesting (and expensive) when we look into self-hosted solutions 🖥. Our NMT (Systran V9) lives on hardware from 2018, and we are happy with its performance. LLM requires a GPU and is not as fast as NMT on comparable hardware + GPU. I would say NMT is at least four times faster, but I don’t have the numbers to prove it. 🏁 TL:DR We love LLMs, and they absolutely have their place as machine translation engines (besides millions of other use cases), but when we need to translate millions of posts per day, we reach for NMT (for now 😃).

5

“domain-specified symbolic engines” To train LLMs on “symbolic engines” for different use cases. What if the “symbolic engine” represented a 3D interior designing software, or a UI designing environment that utilizes “symbols” to built interfaces?

2

LLMs are terrible at basic math. Why? Here's why: Unlike people, LLMs don’t conceptualize mathematics. They simply recognize patterns in the training data (text) they’ve learned from. When asked what 2 + 2 is, an LLM recalls instances of seeing ‘2’, ‘+’, ‘2’ and so on, allowing it to approximate solutions but not match the precision of a calculator. That’s how we end up with LLMs asserting that 2 + 2 = 5. But for analytical teams, like finance, that makes LLMs really hard to adopt in day-to-day workflows. The productivity gains of having AI do calculations are negated by needing to fact-check and audit every calculation and number. LLMs are also not deterministic, so no amount of prompting is going to guarantee you’ll get the correct answer every time. And that’s just for basic math – the outputs I’ve seen from LLMs on more complex calculations and analyses can get very wonky. Add thousands of rows of data and you’re squarely in hallucination land. So what’s the solution? Calculate things manually? No. The good news is that this is changing quickly and there are options: 1 - AI is getting better at math. Researchers continue to train LLMs with more specialized datasets that include a much wider array of mathematical problems and solutions. The newest models, like OpenAI’s latest family of models, are incorporating things like symbolic reasoning and “chain of thought” methods to break complex problems into smaller, logical steps the LLM can more easily process. Results have been really promising and we’ve seen each new iteration of models continue to beat previous benchmarks. 2 - Leverage AI agents For intensive workflows, and to really make sure your outputs are correct, working with companies that are building high-performing AI agents is likely your best bet. Agentic workflows allow LLMs to be used for what they are great at (inferring user intent, making decisions, etc.) while outsourcing the things they struggle with (complex calculations) to other parts of a broader system. Text-to-SQL, “tool-use”, few-shot prompting, and other strategies can be combined with real-time data integrations to dramatically improve performance on analytical operations. At Concourse we’ve spent the better half of 2024 getting our AI Agents to be really good at math, and I’m so excited to continue growing our capabilities to tackle even more complex mathematical problems for our customers in 2025. Reach out if you’re interested in learning more about how we’ve built Concourse and the impact we’re already delivering for finance teams.

42

5 Comments

My wide ranging conversation on enterprise AI automation with Alp Arhan U. on the Masters of Automation is now available on YouTube, Spotify and Apple Podcasts (links in comments). Some of the things we go over are: 🧠 My journey from Turkey to Harvard to YC 🧠 How Brainbase (YC W24) decides what to automate with AI 🧠 Our unique insight at Brainbase 🧠 Why immigrant founders make some of the best entrepreneurs Enjoy our conversation, and please sound out with comments and questions below!

93

6 Comments

After Apple’s research saying LLMs can’t think, it’s funny that now all I’m seeing is research on teaching them to think – something most humans still struggle with. Papers: Looking Inward: Language Models Can Learn About Themselves by Introspection https://lnkd.in/gf6rgnUP On The Planning Abilities of OpenAI's o1 Models: Feasibility, Optimality, and Generalisability https://lnkd.in/gNsPtDbi Thinking LLMs: General Instruction Following with Thought Generation https://lnkd.in/guQtRnsD #AIResearch #MachineLearning #ArticificalIntelligence

24

From my conversations, I'm noticing different trends between how AI engineers deploy evaluation models (model as a judge) in production monitoring vs application testing. 💡 Value - Monitoring: provides real time alerts of performance or used to preemptively block problematic outputs from surfacing to the user -Testing: regression testing applications to establish baseline performance before pushing to deployment ⚡ Speed - Monitoring: near real time, requires extremely fast token processing - Testing: batch, async, often processing large test sets, lower token processing speed required 🎯 Accuracy - Monitoring: overindex on false positives, better to flag something as problematic and be wrong than to let something slip between the cracks - Testing: high correlation with human graders, be as accurate and robust as possible 💰 Cost - Monitoring: prefer cheaper models, otherwise the additional inference cost per generation will add up exponentially - Testing: willing to take on higher costs, calculation is accurate testing leads to better performing outputs 🛑 Risk - Monitoring: if your monitoring judge gets something wrong, there is a high consequence because it directly impacts the end user - Testing: if your testing judge gets something wrong, async review gives some potential to remediate the issue At the moment, it seems using model as a judge for testing is a little more popular due to the high cost constraints of constant monitoring, but that is subject to change as more specialized models come out. As the industry matures, I strongly believe we're gonna see specialization between models for monitoring and models for testing. If you're currently using model as a judge techniques, would love to chat more! Let me know what you think. Twitter Handle: https://lnkd.in/eAsQvtiu

26

Wellfound asked us why a superstar chip designer should join ChipStack. To me, the answer is simple: - The world of designing and verifying chips is undergoing a transformative shift with generative AI. In 2025, we will see its fruition with mature tools. At ChipStack, you can be at the forefront of this revolution. - Anchor your existing skills in design and verification while diving into new territories like AI and enterprise software. Want to grow without leaving behind what you excel at? This is your chance. - If you’ve experienced the inefficiencies of traditional chip design and verification firsthand, why settle for the status quo? Be part of the change! We still have an open role for chip designers: https://lnkd.in/gGZJHNFf. Thanks Wellfound for being an awesome partner!

67