Category mlops | AI | Architecture Knowledge Hub

Building an LLMOps Pipeline: Prompt Versioning, Drift Monitoring, and CI/CD

May 26, 2026

Prompt versioning CLI, LLM-as-judge eval harness, PSI drift monitoring, GitHub Actions CI/CD with a hard quality gate, and a Streamlit monitoring dashboard.

LoyaltyLens Series Recap: Architecture Decisions, Lessons, and What to Build Next

May 21, 2026

Module-by-module architecture decisions, five key lessons from building a production AI system end-to-end, and a prioritised roadmap for extending LoyaltyLens to production.

Wiring Six AI Modules Into One Pipeline — and Deploying to the Cloud

May 19, 2026

How I connected LoyaltyLens's six modules into a single orchestration layer, exported the PyTorch model to TorchScript, and deployed a real-time inference endpoint on AWS SageMaker.

Building a Feedback Loop for Production LLMs: Signals, Preference Data, and Retraining Triggers

May 14, 2026

Closing the production AI loop — feedback capture API, React review UI, preference dataset export, automated retraining triggers, and the practical limits of RLHF without a research team.

#rlhf

#reinforcement-learning

PyTorch Propensity Scoring with TabTransformer

Apr 30, 2026

Building a production-grade offer redemption predictor with PyTorch — TabTransformer architecture, MLflow tracking, model cards, and the SageMaker deployment path.

Building a Production-Grade Feature Pipeline with DuckDB

Apr 28, 2026

Building production-grade ML feature stores without a Databricks cluster — DuckDB, feature versioning, validation gates, and benchmark results.

LoyaltyLens Setup Guide and Glossary

Apr 23, 2026

Complete environment setup, mental model, and plain-English glossary for every acronym in the LoyaltyLens series — RAG, PSI, RLHF, LLMOps, and 33 more.