Nicholas Larus-Stone

This post by Colin Breck feels like it should be one of those that you return to annually in the downtime between Christmas and the new year. Well, for those that take downtime at least. https://lnkd.in/gnndTnX9 One of the topics that resonates with me is the idea of technology selections in the face of uncertainty. I think about it in terms future optionality and minimizing risk; a mashup of concepts from Taleb and Sivers. What decision today is going to give me the most options in the future relative to the downside risk. Note that I didn't say what gives me the most options today, but instead the most options in the future. Too many options today can be paralyzing. Options in the future mean I can be comfortable with a choice today and worry about the added complexity, load, etc. in the future when my application has 100, 1,000, or 10,000 times more users. I'm pretty sure Colin is dead on when he talks about today's two-way doors for platform technologies being object storage and distributed runtimes. It takes time to change your thinking and then change your code but if you are moving in the right direction in consumable chunks I think you'll be happy with the outcome.

2

A question for the software devs out there - do you think that LLMs are to high level languages as high level languages are to assembler ie just another layer of abstraction? Or is the whole field of software development going to go the way of switchboard operators and there will be just be PMs talking to LLMS? 😱

10

49 Comments

"Run your #GitHub Actions locally! Why would you want to do this? Two reasons: - Fast Feedback - Rather than having to commit/push every time you want to test out the changes you are making to your .github/workflows/ files (or for any changes to embedded GitHub actions), you can use act to run the actions locally. The environment variables and filesystem are all configured to match what GitHub provides. - Local Task Runner - I love make. However, I also hate repeating myself. With act, you can use the GitHub Actions defined in your .github/workflows/ to replace your Makefile!" https://lnkd.in/gbtScyq9

2

Do you want to navigate S3 objects as if they're a filesystem? Use AWS Mountpoint. A few things I love: 🔎 ls with tab completion 🎄 tree -L 2 - build a filetree so many levels deep (not possible with aws s3 ls --recursive) 🧭 s3 files and folders appear in notebook file browser 🥽 safety: you can't easily delete files - something I was worried about but they thought about it: by default PUT requests works (cp) but not DELETE (mv and rm) #aws #dataengineering #devex

5

The most downloaded datasets from the California HHS site? If you follow me on LinkedIn, you've probably guessed it - the Medi-Cal enrolled provider list and Suspended and Ineligible list are the top two, followed by other datasets by a wide margin. The California Health & Human Services Agency (Cal HHS) has a nice data hosting site that resembles data.cms.gov. The site looks slick and provides interesting insights, including visitor statistics and download metrics. It's fascinating to see how people are interested in provider eligibility and exclusion data for Medi-Cal billing. Looking at how organizations use these public datasets, compliance emerges as the dominant use case. This makes sense - when accessing government data portals, users are typically verifying facts related to government programs. The user base spans health plans, hospitals, IPAs, medical groups, and even commercial payers, as Medicaid suspensions often trigger reviews across the healthcare ecosystem. What's interesting is that this pattern extends beyond healthcare. On data.ca.gov, which hosts all kinds of government data, the most downloaded dataset is the directory of certified electricians. This shows a consistent theme - whether it's healthcare providers or electricians, people come to government sites primarily to verify professional credentials and eligibility. This raises an intriguing observation about how government data is used. The way organizations use these datasets - downloading lists to verify credentials and eligibility - is functionally similar to what modern Identity Provider (IdP) systems do, but through a much more manual process. What are your thoughts on the future of government data in healthcare identity verification? #Healthcare #DigitalTransformation #Compliance #HealthTech #GovTech #Identity

24

How do you build Athena tables? My #python package Affinity that helps build well-documented datasets and #databases finally has a home! I'm releasing v1 with some enhanced support for #AWS #Athena. The name on #PyPI was taken by an existing project from 2006. The original author was findable, and kind enough to transfer the name to me. There's a formal process to request name transfer through issues on PyPI's github, but it appears to be not functional - crickets in mine and others' posts for months. https://lnkd.in/gkWzGafY

7

1 Comment

(industry) structure prediction: biotech’s cost per approved drug delivered decreases by an order of magnitude over the next decade, due to model progress As one step towards that, the small but mighty Chai Discovery team has released Chai-1, an open-weights, frontier multimodal molecular structure prediction model, on par with AlphaFold3. Chai-1 was shipped after only 4 months of work — it is not an endpoint but a glimpse of what’s coming. More than just a technical milestone, Chai is a signal of the coming structural shift for pharma. Molecular structure and interaction prediction at scale will reduce the time and cost of drug discovery, shifting risk “left” into silico and away from expensive labs and trials. Lower barrier to entry enables scientists to innovate faster and more successfully. The shift of biotech to model-first is exciting for scientists and for society. I can think of few areas where speed of progress is so important. Congrats Joshua Meier Jack Dent Jacques Boitreaud and team! Technical report, open weights and web server for free experimentation: https://lnkd.in/g7SipS_Y

166

8 Comments

🧩 One thing we've noticed with #VentureIQ technology is how well it can distill and communicate complex technical material 🧩 Whether it's a #deeptech venture, a novel #AI application, or even a patent submission, we can quickly distill the business and strategic implications for investors...and sometimes even the founders/PIs themselves! 🤯 Based on our work so far, the tech can handle a wide array of research areas and technological sectors -- anything from blockchain algorithms to vaccine development. It handles the nuance of the field, the specific terminology, and extracts the key concepts which matter for venture-building. 🤓 To demonstrate, I took a patent I helped write with Eric Laber on reinforcement learning. I ran it through #Ventureiq to build a commercial assessment (key differentiators, target markets). Then for fun I had #NotebookLM make a podcast out of it. I've uploaded the result it here (DM me if you'd like the actual report): https://lnkd.in/eiTejsny This is more than just automation or streamlining operations - it's about unlocking clarity and insight when it was previously unattainable, and the ability to do it over and over again no matter the sector or vertical. Interested in more info? DM me so we can chat.

11

2 Comments

Excited to share what Nithin Parsan and I have been working on at YC! At Reticular (YC F24), we're pioneering the future of explainable AI for biology. Check out the post and our first technical blogpost linked in the comments to learn more about what we're focused on.

89

6 Comments

Reimagining Healthcare Workforce Platfforms: Turning Capital Into Labor The future isn’t just about filling shifts. It’s about scaling labor intelligently, driving better outcomes for systems, staff, and patients alike. Andreessen Horowitz and Alastair (Alex) Rampell reinforced this Vision in a recent post and it applies to the next phase of healthcare workforce evolution. Healthcare workforce marketplaces plus AI are at the tipping point of transformation, evolving into systems of record that don’t just manage labor—they optimize it. Much like EHRs centralized patient care data, these marketplaces can centralize workforce operations, aligning talent supply with clinical demand. But there’s a bigger opportunity here: turning capital into labor. Here’s what I mean. Historically, industries (including healthcare) have focused on enterprise computing tasks—streamlining data entry, compliance, and back-office functions. While valuable, this approach often stops short of transforming the real-world, repetitive processes that make up the bulk of work in healthcare. Think beyond enterprise software. The next frontier lies in leveraging embodied AI and workforce marketplaces to automate and enhance labor-driven processes, transforming capital investments into productive labor outputs. For healthcare, this means turning staffing costs into scalable, dynamic systems that deliver value far beyond filling shifts: 1. Dynamic Labor Matching: Capital invested in marketplace platforms enables real-time matching of labor supply with patient demand, replacing outdated scheduling models with predictive, automated systems. 2. Workflow Automation: From credentialing to compliance and shift bidding, embodied AI can streamline traditionally manual tasks, freeing human resources for higher-value work. 3. Labor as a Strategic Asset: Workforce marketplaces shift the focus from static staffing budgets to proactive labor deployment, ensuring that every dollar invested in labor drives outcomes and efficiency. It is about the billions of repetitive, planned tasks that AI can improve. In healthcare, that means automating everything from staffing logistics to clinical task assignments, creating a horizontal system of engagement for labor (think Zendesk) that integrates seamlessly with vertical, workflow-specific platforms (think Toast for staffing operations). The result? Hundreds of billions in untapped market potential—and a healthcare workforce model that transforms how care is delivered. By turning capital into labor through smarter marketplaces and automation, we don’t just solve staffing challenges; we redefine what’s possible in healthcare operations and ShiftMed is leading the way with health systems moving beyond tired models. #capitalintolabor #systemsofrecord #healthworkforce #aiworkforce #healthsystems #hospitals #cno #hospitalceo #healthcareinnovation

11

2 Comments

I spoke today with computational bio expert Jefferson Parker, PhD (at NullSet Informatics Solutions) for our State of AI in Biotech report. He shared some interesting career advice for those working in AI: 9 out of 10 leaders in biotech want AI in their scientific development, but do not know WHY. 1 out of 10 DO know why. They probably have the best data, and that is a blessed place to be and develop in your career. Always ask why leaders want AI and much more will be revealed. And for leaders, understanding why you want AI (beyond the hype and promises) will lead to a team culture that prioritizes data and let's the best talent flourish. It's win-win.

24

3 Comments

Artificial intelligence (AI) holds immense promise for transforming numerous fields, from healthcare and finance to education and transportation. The potential for AI to contribute to the medical field is particularly compelling. Given its ability to process vast amounts of data and identify complex patterns, AI could play a significant role in accelerating the discovery of treatments for diseases that have previously defied conventional medical approaches, such as cancer and AIDS. By leveraging AI's capabilities, researchers may be able to uncover potential solutions that could lead to groundbreaking advancements in the treatment of these challenging and persistent health issues.ArtificialIntelligence #MedicalResearch #HealthcareInnovation #DataProcessing #DiseaseTreatment #ResearchAdvancements

3

Though benchmarks for medical AI are still imperfect, there's no question that AI will continue to march towards diagnostic excellence. These trends will place greater emphasis on physician-scientists – those who advance medical knowledge via closed loop clinic/lab interfaces, which can then feed back into large-scale AI systems that serve entire populations. This was all part of the OG vision for the learning health system. source: https://lnkd.in/eh2RM72s

110

11 Comments

ESM3 from EvolutionaryScale is available now on Amazon SageMaker -- generate complex multi-domain proteins from scratch, create protein design workflows, and incorporate functional understanding #AWSome #AmazonSageMaker #ESM3

10

A few thoughts from BioIT World's symposium on FAIR data 🤖 It was exciting to see a new focus on making data FAIR for not just humans but also APIs and AI. I hope there will be a new wave of tools to leverage this data--and a positive feedback loop where these tools encourage more data standardization. 🌏 The majority of resources and initiatives are based in Europe. What can America do to incentivize more scientific data transparency? 💰 Is FAIR data economically viable for everyone? When FAIR data is not mandated, scientists may not have the resources (time and money) to make data FAIR. What tools could lower the barrier?

42

3 Comments

Great article by Jesse - I would suggest part of the challenge is that the Internet is big, marketing is hard, and biotech software is (relatively) niche. These are tough problems to overcome no matter how many venture dollars you raise. While I'm sure the Deep Origin and Nitro Bio folks who are linked from the article tried to include as many companies and products as they could in their lists, PipeBio and Enpicom are missing under "-Omics Datasets"; RSpace, eLabFTW, and other members of the ELN Consortium are missing under "ELN"; StackWave's Affinity is missing under ELN, LIMS, Experiment/Protocol Design, Lab Inventory, and Visualization; and our LabInspector software is missing under LabOps.

2

Unpopular opinion: Most biotech companies aren't ready for AI/ML, and it's not because of their data scientists. It's because of their bench scientists. Here's why: AI/ML requires large, consistent datasets. But in most biotechs, every scientist analyzes their assays slightly differently. They use different thresholds, different ways of normalizing data, different ways of identifying outliers — even different formulas for basic calculations. This inconsistency makes it impossible to compare results across experiments, let alone build AI models. "But we have SOPs!" you might say. Sure, for running the experiments. But for analyzing the data? Rarely. The hard truth is that without standardized analysis protocols, your data is useless for AI/ML. And implementing those protocols means taking away some of the autonomy bench scientists are used to. It's a tough pill to swallow, but it's necessary if biotechs want to harness the power of AI. At Sphinx Bio, we're building tools that make it easy to implement and enforce these protocols, without sacrificing the flexibility scientists need for exploratory analysis.

478

55 Comments

This one struck a nerve, but there was some misinterpretation -- I'm not suggesting dry lab teams blame their wet lab counterparts! Rather, science is noisy and difficult, so let's not add additional sources of variation at the analysis steps. Standardize what you can and your AI/ML initiatives will be much better off. This requires better tooling, yes, but it also requires education and teams to work together. If your tools don't bring together your teams, then you probably have the wrong tools. The most successful companies of the next decade will be ones which blend software and science natively.

8

1 Comment