📰 Medium · Machine Learning

Creating an artificial intelligence model is only one part of building a useful AI system. After development, the model must be tested… Continue reading on Medi

The infrastructure you use to prototype is quietly sabotaging your production agents. Here’s how to fix the architectural mismatch. Continue reading on Medium »

Through the example of vLLM and TensorRT-LLM. Continue reading on Medium »

MLOps, short for machine learning operations, is a set of practice employed to make machine learning model development deployable, usable… Continue reading on M

PSI + KS testleri, FastAPI, Prometheus, MLflow ve auto-retrain ile uçtan uca mini MLOps mimarisi Continue reading on Medium »

Training a machine-learning model is only the beginning. The real challenge is getting it into the real world, keeping it reliable, and… Continue reading on Med

This post answers: how does anyone outside your notebook actually use the model you trained? That’s model serving, and it’s the piece that… Continue reading on

Serve Phi-3 through Triton’s vLLM backend while keeping its model paths portable from a development workspace to an S3-backed deployment. Continue reading on Da

Memahami data drift, concept drift, deteksi statistik, dan arsitektur adaptive learning untuk machine learning di production Continue reading on Medium »

AI coding agents are the first contributors that read your entire repository and run shell commands non-interactively on day one. But… Continue reading on Mediu

Data pipelines have always been dumb by design — do exactly this, in exactly this order, and scream if anything moves. In 2026, teams are… Continue reading on M

Why most ML models die quietly in production Continue reading on Medium »

A technical deep dive into Project 7 of my MLOps portfolio, an automated pipeline that retrains, evaluates, and redeploys a model with… Continue reading on Medi

Notes from an engineer building around models Continue reading on Medium »

13,000 options. Here is how to pick the right ones. Continue reading on Medium »

The notebook is where ML is born. A modular Python project — with DVC, MLflow, and FastAPI stitched in — is where it grows up. Continue reading on Medium »

Five battle-tested patterns for building scalable, governed autonomous AI systems at enterprise scale Continue reading on Medium »

The Business Problem Continue reading on Medium »

Most ML tutorials stop the second the model says “87% accuracy.” This one keeps going -through tracking, evaluation, testing, logging, and… Continue reading on

I ran Airflow at home to pull live stock data. Here’s how I did it and what I learned. Continue reading on Analytics Vidhya »

What AWS, Azure, and GCP each get right for MLOps, what they cost, and why your existing stack decides more than any feature list Continue reading on Medium »

A comprehensive, battle-tested guide to implementing CI/CD pipelines for Machine Learning projects — and why they’re fundamentally… Continue reading on Medium »

How an MLOps team used ephemeral Docker environments to safely test a car price prediction pipeline — stage by stage — before touching… Continue reading on Medi

Part 2 of an MLOps End-to-End series — 60 models, fully automated, one Airflow DAG Continue reading on Medium »