Mockapapella

Thank you for this layman explanation

Hey HN, a few years back I came across this talk about how `mov` is turing complete and how this can be hacked into writing programs that run entirely using only `mov` instructions (https://www.youtube.com/watch?v=R7EEoWg6Ekk).

A couple years later I stumbled across an article explaining how to implement a simple http server from scratch (https://bruinsslot.jp/post/simple-http-webserver-in-c/).

Naturally, I tried to combine the two concepts but the AI models weren't good enough at the time and I didn't have the bandwidth for yet another questionably useful rabbit hole. Fast forward to today and o1 was able to implement it with some iteration.

I hope you enjoy this quick weekend project. I tried to make the README as helpful as I could to get it running on your own computer easily, though I probably missed some steps. I also tried to use o1 to review our conversation to add reasoning for various changes that were made between the original server implementation and the one that is `mov` compatible.

Ach, sorry 'bout that. I added a red/green colorblind toggle for generating the graphs.

I like it. We'll put it on the roadmap.

Tokens are approximately 3/4ths of a word. That might sound odd if you're unfamiliar, but in short it has to do with how LLMs break down words to process them. Letters are too granular for sufficient quality, full words don't allow LLMs to produce novel text like "dkjmcf0248375cyu18c7437tr18c237m yt034c8nwey4mtc0r23q9p,a;" (should it need to), so the industry has settled on the middle ground of a part of a word.

Yeah I've heard about the accessibility argument. I tried to mitigate that somewhat with the button designs having a black/while outline (depending on the theme), but I may need/want to add a "high contrast" mode later on.

Thank you! We plan on letting users select their own timeline later -- this was more of a defense for us to make sure we didn't get rate limited and everyone could try it out

I think this is fair criticism for where it's at and mirrors my experience while building the tool. For generative AI at least, the smartest models + a good prompt will waffle stomp our tool in terms of quality.

For example, while testing it on "Founder Mode" there were a couple comments that mentioned something like "I hate founder mode but I really really like this other thing that is the opposite of founder mode..." and then just continues for a couple paragraphs. It classified the comment as positive. While _technically_ true, that wasn't quite the intention.

We think there are some ways around this that can increase the fidelity of these models that won't involve using generative AI. Like you said, doing it that way eats a ton of resources.

Main case right now is to see if people like it (and so that we don't hit Algolia rate limits). You're right, a more useful case might be letting the user customize the time frame.

As for mobile, I am but a humble backend dev, but I agree completely. Will put it on the roadmap. Thank you for your feedback!

Oh wow, was not expecting this level of feedback -- thank you for the video!

Addressing some points you made in the video:

1. Yes, this is definitely a V0.1. Thank you for taking that into account :D

2. Interactive graphs (custom colors, intensity, what trends to show, etc.) were something we thought might be useful but wanted to hear what others had to say before we invest the time into a feature nobody wanted

3. That super slow "Searching..." loading message should be showing updates for when it's scraping comments, when it's classifying, etc. Not sure why it's not. Some searches like "open source" are more likely to have a ton of comments to pull so it'll take longer.

4. Default mode changed to light mode because it looks better :^)

1. Comments have been truncated to 100 characters

2. The graphs now go full screen when they are clicked

3. Most comments will be neutral. This is kind of a quantitative way to see "the silent majority"

There's some nuance that is lost due to it classifying the comment as a whole as opposed to specific aspects of it, but it works for the most part. Thanks for trying it out!

It isn't, just straight text

If that's the case maybe I'm seeing a caching thing on my side. I upsized the droplet some

We're actively taking suggestions from people to improve it. The webserver was probably just restarting with the new suggestions :)

(the other part of the team that built this here)

Honestly makes me pretty happy you called out the theme. I've always enjoyed this style of design and was sad to see that it never picked up steam. I love how it seems to combine a digital Material design with a more physical and real feeling. I'm doing my part to bring it back.

The tool definitely has some kinks in it that we have plans to iron out over time; we just wanted to get it in front of people to see if anybody would even like it. Right now it's just grabbing the first 256 tokens and categorizing on that, and it grabs the first 5000 comments (split over 5 calls) over the past month.

    Location: Wisconsin
    Remote: Yes
    Willing to relocate: No
    Technologies: Kubernetes, Docker, ONNX, PyTorch, FastAPI, Django, Python, JavaScript, Postgres, MySQL, AWS, DigitalOcean, Grafana, Redis/Valkey, RunPod
    Resume: Given upon request
    Email: In profile

LinkedIn: https://www.linkedin.com/in/quintenlisowe/

I've built distributed AI systems that can support 35,000 customers inferencing at 30,000 queries per second using transformers and GPU nodes. Separately, I've built classification systems that inference at 200,000 classifications per second. I've founded some companies before (Humanoid robotics and AI) and have leadership experience from when I was the lead mentor for a robotics team.

Pretty slow. I built a sentiment analysis service (https://classysoftware.io/) and 250M words @ ~384 words per message I’m pushing 5.6 hours to crunch all that data, and even at that I’m pretty sure there are ways to push it lower without sacrificing on accuracy.

    Location: Wisconsin
    Remote: Yes
    Willing to relocate: No
    Technologies: Kubernetes, Docker, ONNX, PyTorch, FastAPI, Django, Python, JavaScript, Postgres, MySQL, AWS, DigitalOcean, Grafana, Redis/Valkey
    Resume: Given upon request
    Email: In profile

LinkedIn: https://www.linkedin.com/in/quintenlisowe/

I'm a platform engineer with expertise in AI inference systems (stateless and RAG based), particularly the MLOps side of things. I've scaled GPU clusters to tens of thousands of requests per second in a k8s environment. I've founded a couple (failed) companies, including a humanoid robotics company and an AI Essay writing company with GPT-2, before OpenAI had even made the first commit to their API. I also have a lot of experience working with webservers like Django and FastAPI.

I have some experience in management through mentoring a high school robotics team for a few years and the previously mentioned companies.

I'm mainly looking for IC roles.

As a sort of "showcase" for some of my technical skills, I recently built and open sourced an all-in-one batteries included AI deployment repository: https://github.com/Mockapapella/batteries-included-ai-deploy....

In a similar vein, I set up a kubernetes cluster that you can check out here that involve things like DataDog, LoadBalancers, and more: https://github.com/Mockapapella/kubernetes-proxy-inference

SEEKING WORK | Wisconsin, US | Remote | MLOps Engineer

Technologies: Kubernetes, Docker, ONNX, PyTorch, FastAPI, Django, Python, JavaScript, Postgres, MySQL, AWS, DigitalOcean, Grafana, Redis/Valkey, HuggingFace Email: In profile

LinkedIn: https://www.linkedin.com/in/quintenlisowe/

I'm a platform engineer with expertise in AI inference systems (stateless and RAG based), particularly the MLOps side of things. I've scaled GPU clusters to tens of thousands of requests per second in a k8s environment. At my first company I built an LLM API before OpenAI had even made the first commit to their API. I've been working with Django and FastAPI for a few years now and have a lot of experience working with them.

I have some experience in management through mentoring a high school robotics team for a few years and the previously mentioned companies.

As a sort of "showcase" for some of my technical skills, I recently built and open sourced an all-in-one batteries included AI deployment repository: https://github.com/Mockapapella/batteries-included-ai-deploy...

If you're looking to get a high performance AI API up and running quickly, I'm your guy.

    Location: Wisconsin
    Remote: Yes
    Willing to relocate: No
    Technologies: Kubernetes, Docker, ONNX, PyTorch, FastAPI, Django, Python, JavaScript, Postgres, MySQL, AWS, DigitalOcean, Grafana, Redis/Valkey
    Resume: Given upon request
    Email: In profile

LinkedIn: https://www.linkedin.com/in/quintenlisowe/

I'm a platform engineer with expertise in AI inference systems (stateless and RAG based), particularly the MLOps side of things. I've scaled GPU clusters to tens of thousands of requests per second in a k8s environment. I've founded a couple (failed) companies, including a humanoid robotics company and an AI Essay writing company with GPT-2, before OpenAI had even made the first commit to their API. I also have a lot of experience working with webservers like Django and FastAPI.

I have some experience in management through mentoring a high school robotics team for a few years and the previously mentioned companies.

I'm mainly looking for contract IC roles, but I'm open to full-time or product management.

As a sort of "showcase" for some of my technical skills, I recently built and open sourced an all-in-one batteries included AI deployment repository: https://github.com/Mockapapella/batteries-included-ai-deploy....

This is really cool -- thanks for posting it! I'll have to skim through it at some point since a lot of my work is in classifications models and mirrors the results you've seen

Would have been at one of their AI days in the last few years

Agreed. I'm pretty sure it's due to all the different verticals coalescing, though I can't say for sure since some of the problems are consistent across different workflows as well.

You've got the data scientists tinkering away with the actual models where half of their stuff is in Jupyter notebooks. You've got the platform engineers sticking these models behind an API. DevOps making sure any model/code updates get propagated to prod in a safe and hassle free manner. Then you've got Infra who needs to make sure you actually have the specialized hardware to deploy these models onto.

Better than sacrificing both

    Location: Wisconsin
    Remote: Yes
    Willing to relocate: No
    Technologies: Kubernetes, Docker, ONNX, PyTorch, FastAPI, Django, Python, JavaScript, Postgres, MySQL, AWS, DigitalOcean, HTML/CSS
    Resume: Given Upon Request
    Email: In profile

I have expertise in AI inference systems (stateless and RAG based), particularly the deployment and scaling of systems to service tens of thousands of queries per second. In my free time I like to write software for my personal projects and go deeper into the computing stack of modern computers.

I'm open to remote positions for ML Engineering or Product Management. OK with either contract or full time.

Additionally I have founded a couple (failed) companies, including a humanoid robotics company, and have experience in IoT/Embedded, computer vision, CAD software, web scraping, webservers, and management (through volunteering and the previously mentioned company).

Writing software that makes all other software more efficient, reliable, and easier to run/read.

I really dislike how most software in large corporations degrades over time and increases in complexity. Some pockets are better than others, and there are momentary improvements to bad software, but entropy eventually sets in.

I would build org-wide ratchet mechanisms that make it easier to do the right thing than the wrong thing.

tl;dr generate documentation based off of the code automatically.

My solution might be more applicable to back end python dev (not sure about frontend or other languages), but where I work we have 2 systems for generating documentation based off of code.

1. OpenAPI specs. All our code uses the generated clients. 2. I built a system that has a step in CI/CD that checks the docstrings of the functions/methods to make sure it matches the mypy-enforced function signature, along with some other regex-filters.

As for more high level interactions between software (like a graph view of connected pieces of code), you would probably have to use an LSP which could get hairy.

As for the confluence part, there's probably a working terminal interface (3rd party tool, haven't tried any) that would let you upload the result of all that.

    Location: Wisconsin
    Remote: Yes
    Willing to relocate: No
    Technologies: Kubernetes, Docker, ONNX, PyTorch, FastAPI, Django, Python, JavaScript, Java, Postgres, MySQL, AWS, DigitalOcean, HTML/CSS
    Resume: Given Upon Request
    Email: In profile

I'm open to remote positions for ML Engineering or Product Management. OK with either contract or full time. I have expertise in AI inference systems (stateless and RAG based), particularly the deployment and scaling of systems to service tens of thousands of queries per second. In my free time I like to write software for my personal projects and go deeper into the computing stack of modern computers. Additionally I have founded a couple (failed) companies, including a humanoid robotics company, and have experience in IoT/Embedded, computer vision, CAD software, web scraping, webservers, and management (through volunteering and the previously mentioned company).

oh shit, you're open sourcing this as well? I'd love to use a similar workflow for some of my projects. Love the idea!

Also you guys should post over on Threads -- a bunch of people over there are really into the idea as well: https://www.threads.net/@mockapapella/post/C5_vLdDP0J1

It’s doomed because of the material choice of their pedals? Come on now, don’t be dramatic