Speak at The Fifth Elephant 2026 Annual Conference

Collaborative filtering breaks for new content — there’s no interaction history to learn from. Embedding-based systems improve on that, but they treat items as isolated vectors, blind to the rich semantic relationships between them: shared topics, overlapping entities, genre proximity, mood adjacency. A content knowledge graph makes those relationships first-class citizens and changes what recommendations become possible.

This talk covers a production system that uses a content knowledge graph to power real-time recommendations — not as an offline batch process, but as a live graph traversal at request time. We’ll cover how content is modeled as a graph (entities, topics, attributes, and the relationships between them), how the graph updates incrementally as new content is published, and how traversal patterns translate into a ranked recommendation list. We’ll also cover where the graph outperforms pure embedding approaches — particularly for cold-start content and for surfacing non-obvious connections — and where it doesn’t, and why.

The real-time constraint is where most of the engineering lives: keeping traversal latency within a serving budget, managing graph freshness, and deciding how much reasoning to do at request time vs. pre-compute. Those tradeoffs are what this talk is really about.

For ML and data engineers building recommendation systems who want an alternative to the standard matrix factorisation or two-tower playbook.

Speaker bio:
Himanshu Aggarwal is a Machine Learning Engineer at Glance, where he builds large-scale AI systems for content discovery and personalization, serving over 250 million users globally. His expertise spans recommender systems, semantic retrieval, knowledge graphs, and large language models, with a strong focus on designing scalable, production-grade architectures.

With experience across research and high-scale consumer platforms, Himanshu works on advancing content understanding and building intelligent systems that enhance how users discover and engage with digital experiences across domains.

PPT Link: https://drive.google.com/file/d/1tD1vxjXd-8vmX5YX7674X8T2U6TYQZxw/view?usp=sharing