- Problem: Inefficient manual screening of massive job data.
- Solution: Built a Python-SQL pipeline using LLMs to automate scoring and matching.
- Value: Reduced evaluation time by 90% while increasing matching accuracy.
- Problem: AI hallucinations and lack of domain-specific knowledge in standard RAG.
- Solution: Implemented GraphRAG with Neo4j and Gemini 2.0 Flash for structured knowledge retrieval.
- Value: High-fidelity responses for complex enterprise data queries.
- Problem: Difficulty in real-time social media data collection due to anti-scraping.
- Solution: Developed a resilient system with anti-detection, Celery workers, and advanced proxy rotation.
- Value: Automated competitive intelligence gathering with 99% uptime.
- Problem: Analyzing linguistic patterns in specialized financial/legal corpora.
- Solution: Created a semantic analysis tool using custom NLP tokenization.
- Value: Accelerated domain-specific terminology research for business documentation.
๐ซ Contact me: mrduc2266@gmail.com