Processing Rescued Meteorological Data

Parsing TXT/Excel formatted data into validated DuckDB databases.

Using LLMs for Data Rescue

Rescuing data from scanned handwritten tables using multimodal LLMs and OCR.

Multilingual Topic Modeling using BERT

Using topic modelling with BERT embeddings to identify topics of Telegram channels in multiple languages.

Ongoing and future ideas...

• Indexing my computer as a vector database and building a RAG application.
• Using a computer vision model to search through images on my disk.
• Counting cars driving the wrong way down my one-way street using a webcam and image processing.