Dr. Richard Bergmair



AI Search Engine for Legal Document Intelligence

Contract type: Remote, freelancer
Client: CredCore Inc.
Role: Senior Search and AI Engineer
Time period: Oct 2024 – Apr 2025
Volume: ~ 1000h

ML & AI Engineering • LLM • Prompt Engineering • OpenAI • GPT-4/4o API • gensim • word2vec • Search Engineering • Solr • Quickwit Tantivy • Programming • Python • Data Engineering • NoSQL • MongoDB • Cloud Computing • Linux • Bash • AWS • EC2 • Azure • cloud infrastructure • Docker • Kubernetes • FastAPI • agile development • MLOps • ML Pipeline Design • Natural Language Processing • RAG • Responsible AI • Text Embeddings • Model Deployment • Semantic Search • Git • GitHub • CI/CD • Argo CD • HTML • CSS • stakeholder management • requirements management • REST

Having built an AI-enhanced collection of credit agreements and other legal documents in the domain of corporate credit, CredCore Inc. contracted me to build a search engine to make these accessible to their team of domain experts. I implemented a data model for Apache Solr and Quickwit Tantivy and Python-based ETL software to transform this data from MongoDB to the desired form.

In a separate project, I set up a proof of concept for an LLM-based “chat with a credit agreement” prototype. It utilized retrieval-augmented generation (RAG), reverse RAG, word2vec and doc2vec embeddings provided by the gensim library, and the “annoy” vector search engine.

==> CredCore