Arthur
Buikis
Senior software engineer · Riga, LV
Backend and platform engineer based in Riga. Around twelve years in Python with Rust (PyO3) for hot paths. Currently working on a media-processing platform.
MarkFlow ↗
Solo-built social-analytics SaaS wired into the Meta and Google ad API…
MyProxy
Mobile-phone-as-proxy service for B2B customers. Multi-device rotation…
Expired Domain Search
Celery pipeline that crawled around 700M expired domains — DNS, WHOIS,…
Eight-ish years of Python backend, lately with a bit of Rust. I'm at a media-processing platform right now, mostly working on performance and how the pipeline handles load. Before that: strange-logic (a SaaS called TDN — a lot of crawling, and a PHP-to-Python migration I pushed through), lethub (a scraping pipeline), and Upwork contracts on and off the whole time.
What I do
- 01Python backend. Rust in the hot spots where it earns it (PyO3, maturin, not everywhere).
- 02ML pipeline plumbing — getting models in, keeping the queue honest when things spike.
- 03Performance work when it matters. Indexes, rewrites, the kind of wins nobody notices.
Selected work
Cases —
still writing
them.
Three shown.
More as I write them up.
Python and Rust in a mature Django codebase.
Brought Rust (PyO3) into a production Python platform — started with a graph-read extension, grew into a critical-path ETA engine that walks the job DAG. Spent most of this year on hot-path latency; wins ran from about 30× to 800× depending on the endpoint. Also indexed a large webhook table with zero downtime, and built the shadow-deploy system that ramps new ML models safely.
- Role
- IC · API-perf lead this quarter
- Stack
- python · rust · pyo3 · django · celery · k8s
- Scope
- ~90 MRs · ongoing
- Outcome
- perf wins · 12 ML clients plumbed in
Case notes · drafting →
Crawling a lot of domains, looking for the interesting ones.
Celery pipeline that ran DNS, WHOIS, screenshots, and tech fingerprinting across expired domains — somewhere around 700M of them over about a year. 43 TB by the end. Separately, pushed the team through a PHP → Python migration and brought ClickHouse in when the analytics load stopped fitting in Postgres.
- Role
- core engineer · later senior
- Stack
- python · celery · clickhouse · playwright · redis
- Scope
- multi-year
- Outcome
- 700M domains · 43 TB · php→python migration
Case notes →
MarkFlow — a social-analytics SaaS I built solo.
Product work on the Meta and Google ad APIs — dashboard, ingest, billing. Quietly live at markflow.eu, with paying users. Also earlier: MyProxy (2020–2022), a mobile-phone-as-proxy service for B2B customers. We got the cost per GB down by roughly 20×.
- Role
- solo founder · eng · product
- Stack
- python · typescript · react · meta/google apis · stripe
- Scope
- live product
- Outcome
- shipped · paying users
markflow.eu ↗
[ next case · lethub scraping pipeline — writing it up ]Currently
Building something
on the side.
A small product for operators who want what I want — in a space where the existing options feel clumsy. I'm keeping the details thin while it's still prototype-stage. Full version lives at /building.
Recent updates
- 2026-04-14 · first-run workflow survives a cold boot
- 2026-03-28 · pulled redis pub/sub, put a durable queue in its place