02 / Enrichment & relationship intelligence
MENT
Oct 2023 — Oct 2025AI Engineer · Remote
From scattered data to useful context.
At MENT, my work covered extraction pipelines, data-to-profile attribution, enrichment workflows, and relationship insights. Each addressed a different part of turning source data into useful context.
Pipeline improvement
A shorter path through extraction
I reduced a core AI pipeline from about 15 minutes to under 3. Asynchronous LLM calls extracted different schema sections in parallel, with mapping and validation handled in Python.
Attribution improvement
Matching data to the right person
I improved data-to-profile attribution by approximately 33% compared with the prior system, with particular emphasis on people who had identical or similar names. This is an attribution result, separate from the pipeline timing measurement.
Workflow contribution
Contributing to multi-agent enrichment
I contributed to n8n multi-agent enrichment workflows, including Model Context Protocol tool integration, RabbitMQ integration and retries, and Grafana tracing.
I also built Python microservices and Celery workers for enrichment workloads, and integrated Azure OpenAI into retrieval and conversational services.
Knowledge graph
Relationships that support insights
I designed and implemented a Neo4j relationship graph for shared-interest, employer, and multi-hop network insights.