rmhsilva

Flok Health | Multiple roles | ONSITE | Cambridge UK | Series A

https://www.flok.health

Come help us build explainable AI for providing physiotherapy at scale.

Our core systems are a custom decision engine, interactive streaming video over WebRTC, and multiple internal tools to help us scale. We care a ton about high quality software and great UX - healthcare apps has a bad rep for being unpolished. Commercially, we partner with national healthcare systems to provide high quality care that patients love.

Over the next 18-24 months we are expanding our geographical and product footprint:

- Our infrastructure will need to mature - Our software will need to generalise as we treat more clinical conditions - Our collection of internal tools will need to grow

We're hiring for generalist engineers. Our core stack is Python/Rust/Typescript, but all backgrounds welcome as long as you like ownership and responsibility.

Read more and apply at https://www.flok.health/careers/engineering-team or find me online and DM me! We're 4 days a week on-site in a beautiful farm-style office in Cambridgeshire.

Definitely mention you came from HN if you apply - will help you stand out against the flood of AI-generated applications we get these days.

Flok Health | Engineering (systems/Rust, full-stack/Python/TS, Data) | ONSITE Cambridge UK

Flok builds and runs automated physiotherapy clinics in the UK. Here's an article about us on the BBC: https://www.bbc.co.uk/news/articles/cy4y9xm1eejo

We’re a real clinical service, using a novel DSL, interactive video UI, and other software to scale our patient capacity. We provide treatment for (chronic) back pain, and soon, other common conditions. Flok is both a clinic and a consumer product company, so we can learn fast and perfect the patient experience.

Our core languages: Python, Typescript, Rust.

Read more and apply at https://www.flok.health/careers/engineering-team or find me online and DM me! We're on-site only, in Cambridge.

Flok Health | Engineering (multiple roles) | ONSITE Cambridge UK

Flok builds and runs explainable AI (not LLM-based) physiotherapy clinics. We currently operate in the UK healthcare system.

We’re a real clinical service, built on software solving hard problems across expert systems and real-time video streaming, providing treatment for (chronic) back pain, and soon, other common conditions. Flok is both a "clinic" and a consumer product company, so we can make our patients' experience as good as possible.

Recent article about us on the BBC: https://www.bbc.co.uk/news/articles/cy4y9xm1eejo

We also recently gained class 2a medical device certification for our technology: https://www.digitalhealth.net/2025/10/ai-powered-physiothera...

Our core languages: Python, Typescript, Rust.

Read more and apply at https://www.flok.health/careers or find me online and DM me!

Flok Health | Engineering (multiple roles) | ONSITE Cambridge UK

Flok builds and runs explainable AI (not LLM-based) physiotherapy clinics for the UK National Health System.

We’re a real clinical service, built on software solving hard problems across expert systems and real-time video streaming, providing treatment for (chronic) back pain, and soon, other common conditions. Flok is both a "clinic" and a consumer product company, so we can make our patients' experience as good as possible.

Most recent press about us on the BBC: https://www.bbc.co.uk/news/articles/cy4y9xm1eejo

We also recently gained class 2a medical device certification for our technology: https://www.digitalhealth.net/2025/10/ai-powered-physiothera...

Our favourite languages: Python, Typescript, Rust.

Read more and apply at https://careers.flok.health or find me online and DM me!

Flok Health | Engineering (multiple roles) | ONSITE Cambridge UK

Flok builds and runs AI (but not LLM-based) physiotherapy clinics for the UK National Health System.

We’re a real clinical service, built on software solving hard problems across expert systems and real-time video streaming. Flok is both hospital and a D2C product company, and creating something better than the sum of its parts.

Most recent press about us on the BBC: https://www.bbc.co.uk/news/articles/cy4y9xm1eejo

Read more and apply at https://careers.flok.health or find me online and DM me.

This is great, thanks for the link. Love how it was written in 2004.

Zack (of https://amoveo.io) has a good write up of some attacks here: https://github.com/zack-bitcoin/amoveo-docs/blob/master//oth....

This is fantastic writing, thank you.

I've had a different experience.

In the times where an emotionally charged goal isn't driving me forward, adhering to disciplined regimes for time and knowledge management has helped me to prepare and therefore perform better in times there IS a highly motivating goal.

There don't have to be two extremes of X-management (one fuelled by emotion and one dragged down by apathy). Like anything, building a discipline in the slow times will set you up really well for when you need to run hard in the fast times.

All that said, I use Roam Research[0] for my knowledge management now. I consolidate my thoughts and ideas weekly, and aim for evergeen knowledge[1].

[0]: https://roamresearch.com/

[1]: https://notes.andymatuschak.org/z4SDCZQeRo4xFEQ8H4qrSqd68ucp...

Edit: formatting

Ahh this was a great great article back in the day. Lots of fun on smashthestack.

See also:

- Standard x86 calling conventions -- http://unixwiz.net/techtips/win32-callconv-asm.html

- A x86 instruction reference -- http://ref.x86asm.net/

EDIT: formatting

(OP) Thoughts? Maybe this is totally obvious to some people, but it feels like I only just realised this.

Great to hear you think so -- that's exactly where it's aimed. I'd love to chat more if you're free some time (https://www.linkedin.com/in/rmhsilva/).

Those are great, thanks for sharing.

Fair point.

The killer feature isn't the language; it's the virtual machine abstraction, which lets us stop thinking about the underlying infrastructure and therefore build stuff faster and have simple performance/observability/cost metrics (because the application logic is fully and formally described, rather than spread across many services/infrastructure). Does that explanation make sense?

From a dev point of view, it's this: The ability to take business-logic "units" in Python, and quickly connect them into larger programs/workflows that run on Lambda.

So the language features aren't particularly interesting - it's not meant to replace Python. But it can do things that are hard/awkward using just Python.

One of the FAQs is - can I use it with JS? Does including this sentence not demonstrate the intention to serve a community wider than just Python?

You're totally right. It's much easier to name something without thinking about it - my bad!

Yeah - my bad - will rename this week.

Thanks, that's great feedback!

- "serverless language". Fair point; I'm still working on how to describe it. I'm trying to get across the idea that it's a language for describing serverless applications (rather than a language that runs in a serverless env).

- flow charts: working on it

- new language: you're right, that section is pretty cringy

Hey HN! This is what I've been working on for the last 6 months. I'd love your feedback!

The current goal is to build data pipelines really fast and operate them really well, but I'm also interested to see how far the VM abstraction can be pushed. What would "soft infrastructure" look like?

Cambridge Medical Robotics | Embedded/Software Engineer | Cambridge, UK | ONSITE | http://cmedrobotics.com/careers

Cambridge Medical Robotics is a private company developing the next-generation universal robotic system for minimal access surgery. Our vision is to make keyhole surgery universally accessible and affordable by significantly expanding the range of procedures that can be performed robotically.

We are looking for (lots of) highly capable individuals, with a desire to work in an agile start-up environment to help us achieve this vision.

Get in touch at careers@cmedrobotics.com

Not explicitly stated anywhere on the page, so let's be clear - this will only do 2 layer PCBs (top/bottom copper), correct? Also, what can it do without any "addons"? It looks like an addon is needed to do anything useful (etching, drilling, pick&place). What am I missing?

I reckon anyone willing to spend ~€3000 on hardware will be designing significantly more complicated PCBs than this will be capable of making.

Cambridge Medical Robotics | Embedded/Software Engineer | Cambridge, UK | ONSITE | http://cmedrobotics.com/careers

Cambridge Medical Robotics is a private company developing the next-generation universal robotic system for minimal access surgery. Our vision is to make keyhole surgery universally accessible and affordable by significantly expanding the range of procedures that can be performed robotically.

We are looking for (lots of) highly capable individuals, with a desire to work in an agile start-up environment to help us achieve this vision.

To get in touch: careers@cmedrobotics.com

I haven't had any issues with SyncThing so far, other than the initial setup faff! I haven't had to deal with any horrible merges, and as long as you take frequent snapshots you should be able to recover if the worst happens.

I have a few principles:

- Spurn ~. It generally gets cluttered with machine specific stuff.

- The primary goal is to preserve knowledge.

- The data should be easily readable in 20 years time on hw/sw not yet invented.

My 'digital life' goes in its own folder (eg /foobar), which is Syncthing'ed around to various machines, and backed up periodically. It's very big, and contains a snapshot of 'everything' I want to preserve.

Cron run various scripts to pull data from multiple services (e.g. Pinboard, DayOne, etc) so if they ever go down, I have the data I created on them.

I segment documents and projects that I create / work on from those I've just downloaded to use. This results in a fairly simple top level folder structure:

- docs: Documents and textual data

- dev: Projects, design work, development etc

- external: External tools, software, etc

- media: mostly video/audio

Projects of course do live in their own git repos.

Tools:

- Emacs org mode for notes, todos, etc (best in class)

- DayOne for journaling and notes while mobile (integrated with org notes)

- Pinboard bookmarks

I'm not a huge fan of using fancy project management software tools, because fashions change, tools go out of date, data gets lost etc. Straightforward text files for most things is the best combination of usability and persistence for me. Wikis are great, but suffer from this - they need maintaining, software needs updating, the database format could change, etc. A bit of work to get a text based system going is totally worth it ;).

> The key for me is to remove the psychological barriers to starting a particular task.

Agreed. And one solution is, as you say, writing down a concrete action. The task is suddenly something you can hold in your head, and easily completely mentally.

Procrastination, at least for me, happens whenever I can't see an end to the task, usually because I haven't properly defined the task. As soon as there is a well defined set of steps to take, it suddenly becomes much easier to not procrastinate.

Procrastination is my brain's way of telling me it's not happy with the plan I've made, and that the goals I've set need to be re-written to be more achievable (in particular, more SMART [1]).

That said, projects like this are great, and definitely help with enumerating and planning!

[1] https://www.projectsmart.co.uk/smart-goals.php

Regarding 1):

Could someone with more expertise explain how Keybase protects against MitM attacks please? Does it simply rely on the difficulty of compromising the SSL certs of multiple assertions (twitter.com, github.com, etc)?

As far as I can tell, if someone was able to 'pretend' to be Twitter, ie, MitM an HTTPS connection to twitter.com, they could 'pretend' to be someone who only has their Keybase info on Twitter. Of course, putting your key data in more places makes it harder to appear as you.

Absolutely, web fonts are overrated. For most websites (i.e. websites where the majority of your readers have no idea what the difference between a font and a typeface is), your best bet in terms of speed and cross-platform aesthetics is to use a stack of system fonts.

See http://www.awayback.com/revised-font-stack/ for examples of well chosen cross platform font stacks that each achieve a certain style.

These days you may as well use a 32bit micro, in terms of cost, toolchain and silicon ability.

The extra dollar you spend on those 16bits easily pays for itself in the extra features you often get with modern 32bit micros.

Especially something like this where you might want to do relatively complex maths / algorithms.

e.g. see the stm32 family

Particularly good to see 802.15.4 link layer encryption (AES128) and seamless IPv4 <> IPv6 translation.