Joe Lau

Join us in Irvine on July 17th at 6:30 PM for our latest OC ACM talk, "Systems and Software for Averting an Energy Crisis" presented by Bill Gervasi, a Principal Systems Architect for Wolley Inc. For more info or to RSVP click here: https://lnkd.in/gquYQ2z9 According to a recent Forbes article, "The rise of generative AI and surging GPU shipments is causing data centers to scale from tens of thousands to 100,000-plus accelerators, shifting the emphasis to power as a mission-critical problem to solve." making this talk particularly relevant. #acm #techtalk #gpu #ai #generativeai #nvidia #energy #energycrisis

3

2 Comments

Having lived in SF bay area for so long, I came to know the big coffee culture scene here. I was totally consumed by it a while ago and spent years trying to attain the “perfect" latte. I published a fun coffee story and my entire home setup, which took weeks of research and multiple iterations, in my blog link in bio. Meanwhile, to all fellow coffee loving tech peeps, here are some top tips that no one told me. Put it in practice at home or in microkitchens. - A big factor in preparing milk suitable for latte art is pressure of the wand. If this is important to you, you need to get a machine that has a strong enough pressure. - When pulling shots, put in 18g of beans and extract 30g of espresso. - The darker the roast, the coarser the setting should be on the grinder, the lighter the roast, the finer the setting should be. - When adjusting the grind setting, do so while the grinder is working if there are beans in the hopper. Otherwise, a bean can get stuck between the burrrs. - Pulling a shot should take 20-30 sec in time, or based on observation. It should look like caramel going out, not a waterfall. - Eventually work your way to using a bottomless portafilter. It punishes you for all the small imperfections when preparing your coffee puck, like any tilt, and uneven distribution. It does so by spraying the espresso all over the place. The more you can get it to flow uniformly, the better your preparation of espresso, objectively. - Seal your beans in a vacuum, so they don’t go stale. The more stale the beans, the more minor adjustment you need to make to grind finer. - When steaming milk in your pitcher, you can tell if it’s done by holding the pitcher and as soon as the heat becomes unbearable i.e. stings, it’s done. - Taking a latte art course at a local coffee shop was the best boost for me to start nailing latte art.

9

Included in this article is the mention of vector databases. Unsurprisingly, Michael Stonebraker and Andy Pavlo feel there is not a need for vector-specific databases. They believe relational databases will adopt their features, which is already happening. Check out the details here: https://lnkd.in/gyq_4WjP

27

1 Comment

Anthropic released another paper on their ongoing work towards interpretability of large foundational models like their Claude 3 family, built on previous work by many. Their work continues to be enlightening and inspiring. The findings in this work are generally similar to previous findings on smaller models but scaled up to Claude 3 Sonnet: 💡 they find large numbers of "features" corresponding to combinations of neuron activations and these features map to concepts that we recognize/understand: for example, features for transit infrastructure, particular people, concepts (like bias and "objects lacking sentience"). This might sound familiar from classical machine learning, but the interesting thing is that there is no single thing/neuron encoding each of these concepts, it's a particular combination of neuron activations that embeds these concepts in "superposition" in the network. 💡they observe a sort of "distance" of these feature activations that arrange themselves "near" semantically similar features. When they use their sparse autoencoder approach to learn more and more features they observe that the level of abstraction in the detected features becomes more granular, in ways that are interpretable and meaningful. 💡when they artificially fire these activation patterns during inference, it affects the output in ways that relate to the "feature". For example, asking Claude "what's the most interesting science" normally it responds Physics, but if you clamp the feature activations for the feature "Brain Science" to 10x their max values then Claude responds: Neuroscience. Ways to interpret _and_ influence the model. Great readable work [1] along with an interactive explorer of some of the "features" themselves [2] 🎂 There's an easter egg too: if you scroll in the feature explorer tool and try and select one of the "Randomly selected features from 1M" you get a message "if you want to see the rest of the features, we're hiring!" with a link to their recruiting page :) [1] https://lnkd.in/gZbparwp [2] https://lnkd.in/gC9eus53

19

Moderated a session this morning on applied GenAI/LLMs for a broad group of folks, thanks to Hydrolix.io for sponsoring the event outside RSA. As many folks think about building out "chain of thought" RAG pipelines, I wanted to mention there was a cool effort by a group to expand the context window of LLAMA 3 from 8k to 1M tokens. https://lnkd.in/eK8U3Pms I started to test with a typical needle in a haystack test, but discovered we need a better way to test efficacy- that more accurately aligns with the way we retrieve data from files and content blocks provided to the model. I will share more as a solution is fully identified. BUT, this Gradient model is a great potential addition to those using GenAI Assistant and Agents with local enterprise data, such as large log datasets housed in systems such as Hydrolix.io.

7

Think of your prompts with Copilot as a conversation. Watch this quick video to learn smart techniques for refining your prompts and guiding Copilot to give you just what you need. I like the recommendations that being polite to Copilot will improve results. After all its a conversation and not a one-time demand. #msftadvocate #ai #copilot https://lnkd.in/gm6ViP-p

4

1 Comment

Here's my article on how to avoid R&D tax incentive landmines, first published in Startup Daily - https://lnkd.in/gSDv8YeT #tech #founders #technology #startups

16

1 Comment

Who uses Markdown anyway?? That's right, more than you would imagine! The next write up on my list was Markdown itself. Markdown is the heart of Pipedown: https://core.pipedown.dev. Markdown is: simple, accessible, used widely and has a massive ecosystem built around it. To get to grips with the breadth of tools we can leverage I've put together a little article detailing some of the cool things you can do with Markdown and how I imagine they will intersect with the Pipedown workflow. https://lnkd.in/dtbtG4Rj As a happy side effect, this kind of research helps me hone the use case for Pipedown. There's a vast amount of prior art to build upon and leverage. No need to reinvent. I have spent a lot of time experimenting with how Pipedown itself can serve Markdown files. It works fine. However, Static Site Generation and hosting is a solved problem! I can just piggyback on one of the many tools available and concern myself with how Pipedown can enhance and work with a tool like: https://lume.land/

4

Who was Jim Simons, and why should you care? James Simons(1938-2024) was a mathematician, the founder of Renaissance Technologies- commonly referred to as RenTech. He died 10th of May, 2024. He pioneered quant, or quantitative analysis, being one of the first people to apply heavy data mining and using computers for data processing to generate tremendous returns on investments. RenTech generates anywhere between 32-66% returns on investment annually. There's something for everyone in his story. 👉Jim Simons founded RenTech when he was around 44 years old. Talk about a middle aged mathematician starting a hedge fund. 👉 He was an accomplished mathematician before RenTech, having worked on geometry of manifolds. I'll admit I don't have a deeper understanding of his work here 😬 👉 He rose as high as #23 on Forbes' list of the wealthiest people in the world being one of the highest paid hedge fund managers. 👉 The Simons Foundation funded Arxiv, the much beloved non-profit preprint research server. 👉 His foundation also funds the amazing Numberphile channel on YouTube, and worked a lot for mathematics education. 👉 He gave very few talks or public appearances, but whenever he did it was highly informative - check out his Numberphile video. 👉 There's a book about him and his work- The Man Who Solved The Markets by Gregory Zuckerman. If this post piqued your interest, you should definitely check it out.

41

3 Comments

CambioML is taking a bold step: We're making the AnyParser API FREE! Generate your self-service API key at https://lnkd.in/gCUt4QRQ #parsing #extraction #pdf #ppt #word #document #genai #llms

34

1 Comment

I've got some time freeing up for consulting/advisory work, starting in a few weeks -- I primarily work with what I'm calling 'busy teams' -- SREs, Dev[Sec]Ops, Production Engineering, Incident Command, IT, and so on. Extra bonus points if you've a Dublin office, but not entirely necessary :-) I am timezone-flexible and can work with semi-regular travel. The work I've been doing with 'busy teams' amounts in roughly equal measure to: - Org and mandate structure work (Why is the team here? How can we structure for the work? how do we recover from staffing woes? and so forth...)  - Technical and directional work (Platform Modernisation, Cloud Strategy, SLOs, Fractional CxO/PE.)  - "Invisible Hand" style group leadership coaching and advice, embedded with the leadership team or otherwise.  - Direct intervention, including 'sharding' teams and building out new office or geo presences (Dublin preferred for this one, of course).    That's a pretty broad brush - if you could use a *deep inhale* world-class SRE/Production/IT/Ops leadership and technical practitioner to work with you on what you'd like to do, drop me a line on here, or schedule some time via Calendly at https://lnkd.in/efvftJbM If you'd like a lovely packaged version, including a portfolio of previous work, check out https://coservant.systems/ -- Friends, please feel free to reshare if you feel the world needs more daveoc. Thanks!

14

To commit, according to definition = to deliver. I was just curious. https://lnkd.in/gxqhZ3Nr

3

I have a new article about how to use OpenTelemetry in Event Driven Architectures (more specifically with Kafka and NestJS): - Medium: https://lnkd.in/gA8UgPmE - Hashnode: https://lnkd.in/gHmAReNj

6

Seems important to post here, since this continues to be the largest existential threat to humanity in my estimation: https://lnkd.in/gSKv875R I suppose there are two ways to read the departure of the entire "alignment" / AI safety team: - OpenAI doesn't really care about safety, because achieving AGI is their goal regardless of the consequences - AGI alignment is effectively impossible, and the researchers are just burning cycles for a fundamentally intractable problem I'd guess the latter is accurate, although it sounds like the former is more the cause for the departures.

1

On Agent Dev & The Case for The Engineer’s Creative Process Agent development pulls engineers closer to the headspace of artist - and my bet is if engineers embrace and protect their creative processes, we unlock the next level of innovation in this agentic era. In machine learning, the saying goes, "it is more art than science." Building agents is a non-deterministic process that adds significant variance into systems. Engineers would be well-served to hone their creative process. The best engineers will become artists. 1. Building isn’t linear. Building agent infrastructure is far from linear. Engineers often experience a cycle of frustration and eventual breakthroughs. While not perfect, these reasoning systems are now stable enough for integration into our tools and lives. The history of AI, from machine learning to deep learning, has always involved non-linear development. Today's agent infrastructure development is even less linear, requiring engineers to understand complex semantic translations and carefully manage user rollouts to avoid failures. 2. Building with a relationship to the work. With new forms of intelligence come new relationships. Engineers must navigate a relationship to work that involves agent memory and understanding. What is it to build a system that knows you? When people feel understood, they feel loved. What is it to love the systems we’re creating? How does a creative process help? A non-linear process and a relationship to the work are the territories of the artist. Engineers, trained towards control, must embrace uncertainty, risk embarrassment, and accept that their efforts might yield unexpected results. We’re glimpsing a creator's role in a system that understands. While engineers have largely operated in a logical space, they are now called to form a heart connection with the systems they build. A creative process helps artists endure challenges and breathe life into their work. It involves a commitment to creating, discipline, and accepting that one can only control the process, not the outcome. Agent development pulls engineers closer to the mindset of artists. If engineers embrace and protect their creative processes, we can unlock the next level of innovation in society. Read more on this on my Twitter: https://x.com/ashebytes

16

2 Comments

A data-driven approach to driving wound healing outcomes Marcella Gomez, University of California, Santa Cruz https://lnkd.in/gAkWZYJT Meet: https://lnkd.in/g4AAWkcK YouTube Stream: https://lnkd.in/gwtqaTmf Talk details: https://lnkd.in/gWn3B3cU Subscribe for talk announcements: https://lnkd.in/g5ciuNuX Abstract: Precision medicine requires an ability to predict the response of an individual to a prescribed treatment regimen a priori. Thus, advancement in the field is challenged by a lack of predictive models and, arguably, a lack of time-series information for a highly dynamic system. Here, we discuss work in wound healing, for accelerating wound closure. We argue that timing of treatments is as critical to consider as the choice of drug or therapy. Due to system size and complexity, data-driven methods need to be explored to develop multi-dimensional quantifiable indicators tracking systemic changes. In this work I discuss how bioelectronic devices enhanced with deep learning can help facilitate real-time sensing and actuation for automated decisions in treatment for wound healing and preliminary work in transcriptomic based classification of wound states. Bio: Marcella M. Gomez is an associate professor at UC Santa Cruz in the department of Applied Mathematics and Associate Dean for Diversity, Equity, and Inclusion for Baskin Engineering. She received her PhD from Caltech in 2015 and a B.S. from UC Berkeley in 2009; both degrees in Mechanical Engineering. Her research interests are in the broad field of bio-control leveraging methods in machine learning and control theory. Applications range from controlling single-cell response to driving complex systems such as wound healing. #modeling #simulation #medicine #healing #trauma #predictivemodeling #datascience #data #woundhealing #biology #bioengineering #systemanalysis #sensing #omics

5

Verticalmove, Inc is hiring several Senior and Lead level Engineers for Generative AI initiatives within an established billion-dollar client located in the Seattle area. [PLEASE NO C2C INQUIRIES. IF YOU ARE A C2C FIRM AND REACH OUT, YOU WILL BE BLOCKED] Innovation Laboratory that’s embedded within an established Fortune 50 client. This innovation lab is responsible for the future of Augmented Reality, Virtual Reality, Robotics, and Generative AI solutions for its parent organization. These roles are perfect for engineering pros who like use words like INVENT, INNOVATE, and DISRUPT: • Senior Software Engineer for the GenAI team, specializing in AI technologies such as Computer Vision and Machine Learning. Working on the Smart Living Domain, building a virtual assistant to process and organize data for Language Models (LLMs). Responsibilities include fine-tuning models, optimizing LLMs, and integrating them into a serving interface. • Senior Software Engineer with expertise in all aspects of data, including processing and pipelining. Focused on making data accessible and easily searchable, while exploring future enhancements through intelligence integration. Interested? You must be willing to work on-site 50% of the time in their office located on the Eastside of the Greater Seattle area (Bellevue, Redmond, or Kirkland). Please get in touch and say hello@verticalmove.com #MachineLearning #DeepLearning #AI #ArtificialIntelligence #ComputerVision #GenerativeAI #DataScience #BigData #TechInnovation #DataAnalytics #AIResearch #AIInnovation #MLAlgorithms #CVAlgorithms #DataDriven #FutureOfTech #AITransformation #TechTrends #DataEngineering #SmartTech #ComputerVision

6