Model Serving Fundamentals & REST APIs
A trained model sitting in a .pkl file helps nobody. Today your iris classifier gets a front door: a REST API that any application, in any language, can call over HTTP.
Sandip Das
Senior Cloud, DevOps, MLOps & ML Platform Engineer | Heading Cloud, DevOps & MLOps for start-ups | AWS Container Hero | Educator | Mentor | Teaching Cloud, DevOps & Programming in Simple Way
Check out the latest posts
Review Week 2 & The Containerized ML Experiment Workflow
Two weeks in, you have all the pieces: Git and DVC for versioning, MLflow for tracking, and Docker for reproducible environments. A single docker compose up that spins up a tracking server, runs a versioned experiment, and logs every result.
Docker Compose for Multi-Service ML Environments
Stop running MLflow and your training script in separate terminals. Today you'll wire them together with Docker Compose , one command, a full local ML platform, ready to log every experiment automatically.
Writing Production-Grade Dockerfiles for ML
Multi-stage builds, .dockerignore hygiene, layer caching, and dependency pinning , the engineering discipline that separates a "it works" image from one you'd actually ship to production.
Docker for ML Engineers
Containerize your ML code so it runs identically everywhere — on your laptop, on a teammate's machine, and in production — with zero "works on my machine" surprises.
Hyperparameter Tuning & Experiment Comparison
Systematic search strategies, MLflow-tracked runs, and the art of comparing hundreds of experiments to extract your best-performing model.