Posts

Complete MLOps Series: ← Part 1: Pipeline | ← Part 2: Deployment | Part 3 (current) Anatomy of an MLOps Pipeline - Part 3: Production and Best Practices 11. Model and Parameter Selection Strategies The Complete Flow: Selection → Sweep → Registration This pipeline implements a three-phase strategy for model optimization, each with a specific purpose: Step 05: Model Selection ├── Compares 5 algorithms with basic GridSearch (5-10 combos/model) ├── Objective: Identify best model family (Random Forest vs Gradient Boosting vs ...) ├── Primary metric: MAPE (Mean Absolute Percentage Error) └── Output: Best algorithm + initial parameters Step 06: Hyperparameter Sweep ├── Optimizes ONLY the best algorithm from Step 05 ├── Bayesian optimization with 50+ runs (exhaustive search space) ├── Objective: Find optimal configuration of best model ├── Primary metric: wMAPE (Weighted MAPE, less biased) └── Output: best_params.yaml with optimal hyperparameters Step 07: Model Registration ├── Trains final model with parameters from Step 06 ├── Registers in MLflow Model Registry with rich metadata ├── Transitions to stage (Staging/Production) └── Output: Versioned model ready for deployment Why three separate steps? You don’t have computational resources to do exhaustive sweep of 5 algorithms × 50 combinations = 250 training runs. First decide strategy (which algorithm), then tactics (which hyperparameters). ...

Post Objective Data cleaning and preliminary analysis process Understanding the emotional charge or plot development of texts through semantic archaeology based on PCAs Understanding the connections and most representative ideas within the document set Intention Understanding a story’s behavior at the level of its variance is a challenge addressed by attentional engineering. Therefore, using lesser-known methods such as the vector narrative arc combined with a literary map constitutes an interesting route to address increasingly common problems. ...

MLflow for Generative AI Systems I’ll start this post by recalling what Hayen said in her book Designing Machine Learning Systems (2022): ‘Systems are meant to learn’. This statement reflects a simple fact: today, LLMs and to a lesser extent vision language models are winning in the Data Science world. But how do we measure this learning? RLHF work is always a good indicator that perplexity will improve, but let’s return to a key point: LLMs must work as a system, therefore debugging is important, and that’s where the necessary tool for every Data Scientist, AI Engineer, ML Engineer, and MLOps Engineer comes in: MLflow. ...

Reflections on TensorFlow in the context of the modern AI engine and the evolving role of Data Scientists Throughout my journey in the world of data, I’ve witnessed many changes — some tools fading out of popularity while others take the spotlight. R, for example, has become more niche, used mostly by statisticians and academics. Flask, once a common choice for lightweight APIs, gradually gave way to FastAPI thanks to its modularity and support for asynchronous features, redefining how APIs are designed and deployed. ...

Less than two months ago, the most powerful version of the Raspberry Pi 5 hit the market, featuring 16GB of RAM. While its price ($120 USD) is a valid discussion point, as someone who uses these devices as servers for deployment testing and efficiency evaluation at the code level, I want to explore its utility from a computer science perspective in the context of MLOps and LLMs testing. Raspberry Pi Utility Let’s start with some common applications to build on ideas: ...