When a software engineer faces the challenge of organising 3.5 million products for millions of monthly users, the problem is no longer just about writing code. It is about high-level system architecture. For Anslem Chibuike, this challenge became a career-defining project that recently gained international recognition.
Anslem, a Software Engineer who recently saw his technical approach featured as a global case study by ZenML, has become a standout voice in the conversation around large-scale data orchestration. His work focuses on a problem many engineers fear: high-dimensional data at enterprise scale.
The Challenge of 350 Million Data Points
The core of Anslem’s work involved solving a massive search relevance bottleneck for Zoro, a subsidiary of the Fortune 500 company Grainger. In large-scale industrial e-commerce, a product is not just a simple entry. It is a collection of up to 100 unique technical attributes, such as thread size, material grade, and pressure ratings.
“When you scale those attributes across 3.5 million products, you are effectively managing over 350 million data points,” Anslem explains. “Traditional sorting methods simply cannot handle that level of complexity without the system collapsing under its own weight.”
The AI Frontier: Balancing Innovation and Accuracy
What set Anslem’s approach apart was his decision to introduce Artificial Intelligence, which was still a relatively unproven tool at this scale.
“This was one of the very first integrations of an AI approach to solve a large-scale business problem of this nature,” Anslem notes. “Using AI in our workflow at this scale was a first for us, and it definitely kept me on my toes,” Anslem explains. “I was constantly walking a line between the AI speed and the absolute need for facts. AI can sometimes be a bit too creative, what we call hallucinations, and in a business selling industrial parts, you cannot have the system guessing a spec or a measurement. My job was to engineer this scale, so we always had the option to intervene. I built validation layers and checks so that if the AI was ever uncertain, a human could step in and verify the result before it hit the live site.
Engineering for Millions of Users
For a platform like Zoro, which serves millions of monthly visitors, performance is the ultimate metric. Anslem knew that even the most accurate sorting logic would fail if it introduced latency.
“In a high-traffic environment, you don’t have the luxury of slow processing,” says Anslem. “The challenge was to provide a perfectly logical order for the user without adding a single millisecond of delay to the page load. You have to balance deep mathematical logic with extreme performance.”
To achieve this, Anslem shifted the complexity away from the live search environment and into a dedicated data pipeline. He developed a two-stage strategy:
The Normalisation Engine: He built a system to standardise inconsistent data from thousands of sources, turning messy technical specs into a unified mathematical scale.
Pre-calculated Weighted Scoring: By engineering the system to calculate relevance scores during data ingestion, Anslem ensured the heavy lifting was completed before a user ever performed a search.
A Global Technical Benchmark
The result was a high-performance architecture that delivers instant, relevant results across a massive catalogue. This implementation was so effective that ZenML, a global leader in MLOps, selected Anslem’s work for documentation as an official technical case study.
For the tech ecosystem, Anslem Chibuike’s work serves as a blueprint for scaling. It demonstrates how a Software Engineer can transform a chaotic data environment into a high-performance, global standard system. By focusing on the intersection of data integrity and system speed, Anslem has set a new benchmark for handling high-dimensional data in the modern tech
