Essays and technical notes on system design, product engineering decisions, and practical trade-offs.
FEATURED
Featured
May 19, 20268 min read
Modern AI systems are becoming increasingly orchestration-heavy, but optimizing them based only on final outputs is becoming unreliable. This article explores how multimodal workflows, model variability, latency, token costs, and debugging opacity pushed me toward thinking more about inspectability, workflow visibility, and AI systems engineering.
Featured
May 19, 20268 min read
A research-driven exploration of why semantic similarity alone is insufficient for AI memory systems, and how retrieval quality depends on ranking policies, recent context, hybrid retrieval, and adaptive query-aware strategies.
Featured
April 10, 20265 min read
I built a multimodal emotion recognition system using audio and text, expecting combined models to outperform single modalities. In practice, simpler models and selective feature extraction performed more reliably than complex integrations.
ALL ARTICLES
April 10, 202610 min read
An end-to-end image-based recommendation system that combines classification, feature extraction, and similarity ranking to generate meaningful clothing recommendations from user-uploaded images.