JobLeads

JobLeads matches candidates against tens of millions of job listings, and search has to answer in a few hundred milliseconds, every time, under massive filtering.

I work on the search architecture itself: migrating from Elasticsearch to Qdrant vector search, building large-scale filtering pipelines, and optimizing latency across an autoscaling Kubernetes microservice fleet on AWS.

On top of that sits the intelligence: LLM-powered matching and resume processing, shipped as product through SSR Vue frontends, feature flags, A/B testing, and Mixpanel analytics.