AlmazMurzabekov
|10+ years wrangling big data
I design and run the unglamorous machinery behind analytics and AI — cloud data platforms, streaming pipelines, agents that handle the boring half of ingestion, and the CI that keeps them all honest. Occasionally I write about it. Occasionally in Rust.
staging_agent.py — running
from google.adk.agents import SequentialAgent
# profile → model → dry-run → land
staging = SequentialAgent(
name="dbt_staging",
sub_agents=[
profiler, # sniff schema + types
modeler, # emit staging SQL
checker, # dbt build --empty
])
staging.run(source="raw.sensor_feed",
autoland=True)
✔ 128 models01About
Hi 👋 I make data behave.
I'm a Data Engineer with more than 10 years of experience. I've always been drawn to high-load and big-data systems, and I've been lucky enough to chase that curiosity through both work and side projects.
These days that means cloud-native platforms on GCP — Airflow and dbt for the pipelines, Terraform for the ground they stand on, and multi-agent systems on Vertex AI that take the tedious half of ingestion off my hands. With AI everywhere, the bottleneck moved back to where I live: the data. In my free time you'll find me learning something new — usually Rust, systems design, or how LLMs work from the inside.
10+
years in data
PB
scale pipelines
MSc
Big Data Systems
Certs
GCP Certifications
Education
2019 — 2021MSc, Big Data SystemsHigher School of Economics, Moscow
2011 — 2015BSc, Applied InformaticsSaratov State Technical University
02Projects
What I'm building now.
almaz.murzabekov.net/blog?tag=building-own-llm
Building My Own LLM
Build log/2026 ->
Building a language model from scratch, one component at a time. The notes from the way live on the blog.
03Contact
Got a messy data problem? I like those.
Consulting, collaboration, a podcast, or just an argument about whether the medallion architecture is over-hyped - my inbox is open.