Production2023-2024Research Infrastructure

Researchers Search for UNITE!

Semantic Search & RAG for European Research Network

Built an intelligent research discovery platform serving 2000+ researchers across 7 European universities. Combines automated data ingestion, semantic search with hybrid scoring, RAG conversational interface, and interactive UMAP visualizations to accelerate research collaboration and grant team formation.

<500ms

Search Performance

Average semantic search response time

90% faster

Team Assembly Speed

From 3-4 weeks to 2-3 days

4.6/5.0

User Satisfaction

Average rating from coordinators

The Challenge

The UNITE! university alliance needed to enable research discovery across 7 European universities with fragmented data sources, multilingual content, and no centralized search infrastructure.

❌

2000+ researcher profiles scattered across institutional databases

❌

Publication data fragmented across Scopus, Semantic Scholar, and institutional systems

❌

Traditional keyword search missed conceptually related expertise

❌

No automated data ingestion—manual updates were impossible at scale

❌

Multilingual content (English, Polish, German, Spanish, Italian) required cross-lingual matching

❌

No visualization of research networks and collaboration opportunities

The Solution

Designed and built a comprehensive AI-powered platform with four core technical components: (1) Automated data ingestion pipeline scraping academic APIs, (2) Hybrid semantic search combining vector embeddings with keyword matching, (3) RAG conversational interface with source attribution, and (4) UMAP-based network visualization for 2000+ researchers.

Tech Stack

FastAPINext.jsPostgreSQLpgvectorLangChainSentence TransformersOpenAI GPT-4Anthropic ClaudeUMAPSupabase

Need to build intelligent search and discovery for your research or enterprise data?

I specialize in semantic search, RAG systems, and AI-powered knowledge platforms that make complex data discoverable and actionable.

Schedule a Consultation