Skip to content

// blog

Technical writing

Practical, production-grounded notes on building AI and data systems.

Medium RSS
8 min read

Designing a Production Recommendation System That Actually Moves Revenue

A pragmatic walkthrough of building a production recommendation system with candidate generation, propensity ranking, a Kafka/Airflow/BigQuery pipeline, and metrics that track revenue, not just AUC.

  • #recommendation-systems
  • #machine-learning
  • #data-engineering
8 min read

LLM Observability in Production: What to Actually Log and Measure

A pragmatic guide to LLM observability in production: cost, tokens, latency percentiles, failure rates, groundedness, prompt versioning, tracing, evals as monitoring, and alerting.

  • #llm
  • #observability
  • #mlops
9 min read

Real-Time Data Pipelines with Kafka: Lessons from 300 req/min Under 25% CPU

Hard-won lessons from running a real-time third-party data platform on Kafka: partitioning, backpressure, exactly-once vs at-least-once, schema evolution, and circuit breakers.

  • #Kafka
  • #Data Engineering
  • #Python
9 min read

Multi-Armed Bandits in Production: When A/B Testing Wastes Your Traffic

A pragmatic guide to running multi-armed bandits in production: epsilon-greedy vs Thompson Sampling, delayed rewards, non-stationarity, guardrails, and wiring.

  • #multi-armed-bandits
  • #ab-testing
  • #thompson-sampling