🕩 Hello! My name is Sameer and this is my blog.
🧰 Nix (advanced), Infrastructure Orchestration (advanced), Microservices (intermediate), Serverless (intermediate), Kubernetes (intermediate), Distributed Systems (intermediate), Data Warehousing (intermediate), Blockchain (intermediate), Amazon Web Services (intermediate), Rust (intermediate), Haskell (intermediate).
Once proven, I abstract the architecture for scalability using IaC to automate, simplify, and manage stateful components across deployment, distribution, and control.
> Also I find it way less distracting and more productive and learning experience, than training AI assistant or dumping my data to bigtech storage apps, sync which routinly decouple their FOSS api clients in their TOS, for my routine tasks by, how I function. I do use AI sectively for grunt work, eg RTFM, scroll 20k nixos Options, train phone browser tabs, etc. I am more inclined towards federated communites, decentralized chat servers, `openpublish` literature, in contrast to toraliatarian control of resources.
Just like this, I built
A CLI util jobpipe, which given the relevant data, it shortlists role openings for the candidate. Symbolic concepts from a typed ontology and DSL are projected into feature and embedding spaces for probabilistic ranking against user feedback, producing sparse features consumed by the ML training pipeline to re-rank and generate the shortlist.
A commenting service rs-comment-api A Rust microservice deployed as a NixOS systemd service, using IndieAuth for identity, and Turso libSQL for storage - exposed a virtual service via Tailscale Funnel or Cloudflare Zero Trust tunnel. No accounts,no tracking.
Essays
I have written an essay on my pursuit to understand complex systems, given its relevance in the era of AI.
Content and Publishing
Usually, I spend time in defining the apparent problem, which I am trying to solve, then draft the details referential knowledge, with due citations, followed by actually solving an issue and/ or building and testing something useful for it, then deploying it if needed . And finally I update it with working knowldge of the build and a better definition, with due reason to why is this a more precise definition, with due explainations and a contrast between the adoped and notable other approcahes, which help me further understand the probelm and/or reason the approach and may become useful with development, if things begin to break, during applying it further in different possible scenarios. That’s nearly how I ve also experienced the process of education.
Architecture
This blog is built with Hakyll and a Rust/Wasm frontend, deployed via GitLab Pages. It previously deployed to SourceHut Pages with a simple bash script:
#!/usr/bin/env bash
cd dist
git ls-files --others --ignored --exclude-standard -z | xargs -0 -n1 dirname | sort -u > ../exclude.txt
cd ..
tar -C dist -cvz -X exclude.txt . > site.tar.gz && \
curl --oauth2-bearer "$(pass access-tokens/sr.ht/hakyll)" \
-Fcontent=@site.tar.gz https://pages.sr.ht/publish/carnotweat.srht.site
rm -v site.tar.gz exclude.txtIt now builds on a GitLab runner hosted on a production QEMU NixOS VM running two Kubernetes nodes - no cloud compute, no managed services.
The plan is to grow this into a suite of small, self-hosted web services on the same infrastructure - a full-featured web presence without cloud dependency, just with libsql PAAS.
Contact
If you happen to use any of these or write such tools, or for some specific note
📧 xameer@disroot.org for anything related to this blog, collaboration, or microservice work.