01
Production ML pipelines
Feature engineering, model training and hyperparameter tuning brought together into one config-driven pipeline that runs on its own in CI.
I build automated data and model-training pipelines, make large datasets fast and cheap to work with, and make sure the metrics your team relies on are real.
Feature engineering, model training and hyperparameter tuning brought together into one config-driven pipeline that runs on its own in CI.
Streaming and batch pipelines that turn billions of raw events into clean daily datasets, tuned for speed, memory and storage cost.
Finding data leakage behind scores that look too good, and setting up automated health checks that run after every training.
Models that predict conversion and customer lifetime value early in a user's life, to guide budgets and campaign optimization.
Warehouse tables and pipelines that turn raw product and marketing data into reliable datasets, and the data behind your dashboards.
Working with product and marketing teams to turn open questions into analyses, and analyses into decisions.
Agents that automate routine work on your own data: recurring analysis, reporting and monitoring.
Customer support agents, personal assistants and similar conversational agents, built around your own knowledge and tools.
I use modern AI tools at every stage of this work, from writing and reviewing code to diagnosing models and speeding up analysis, so results reach your team sooner.
Tell me what you are building and where it hurts.
blagovda@gmail.com Telegram: @blagovda