Writing
Engineering Lessons from Production AI
I write about designing, deploying, and operating AI systems in production. My articles focus on the engineering challenges that emerge after a prototype becomes a real product—retrieval quality, evaluation, latency, scalability, monitoring, and reliability.
Current AI Engineering
Recent writing aligned with production LLM, search, and enterprise AI delivery.
Why Vector Search Never Says "No Results"
A production lesson on false positives, confidence thresholds, and why semantic retrieval needs explicit failure handling.
How LLMs Can Benefit from Self-Generated Context: Memento
A practical synthesis of Memento research and why context compression matters for KV cache, throughput, and cost in long-running reasoning systems.
LLM: How Response Generated - Sampling
A practical explanation of sampling controls like temperature, top-k, top-p, and constrained decoding for balancing determinism and creativity.
How LLMs Are Built
An end-to-end overview of LLM architecture, training lifecycle, and application-layer concerns including alignment and decoding.
Eurowings Search Modernization on Databricks
A case-driven view of improving retrieval quality, latency, and customer self-service in airline digital journeys.
Data Science and ML Foundations
Applied machine learning and data science articles that show modeling depth and evaluation rigor.
Internet and Rumors - Identify It Faster
A research-to-practice summary of early rumor detection using dual CNN design and certainty-factor-based activation.
Jupyter Notebook and Kernel - Basics and a Fix
A practical guide to Python environment and kernel mismatches in notebook workflows, with clear debugging patterns for data science teams.
Connect Jupyter with Amazon AWS Athena
A practical walkthrough for querying large historical datasets from notebooks using Athena, MFA, and Python connectors.
A Classification Task Using Theano
A from-first-principles walkthrough of logistic regression internals, optimization, and evaluation loops in a deep learning framework.
Performance Metrics in Machine Learning - Part 2
A practical survey of classification and regression metrics, with guidance on when each metric is informative or potentially misleading.
Engineering Archives
Selected platform and architecture notes from previous engineering domains.
Why and How to Protect Your Web Against Bots
A practical security engineering guide for identifying malicious bot activity and applying layered protections for web and API systems.
Why Session Handling Is Difficult in AEM
A practical deep dive into state management limits in RESTful AEM flows and realistic alternatives for cross-page information transfer.
Connection Pooling - Integration with AEM
A performance-focused guide to HTTP client connection pooling for AEM integrations with external systems.
CSRF in AEM
A practical walkthrough of CSRF token handling, referrer controls, and CORS-related protection in AEM environments.
AEM 6.2 - Issues and Resolution
Field notes on upgrade-related issues, diagnostics, and practical fixes encountered while adopting AEM 6.2.