AI and ETL for recycling plant design

Description

Humara has software for the design of recycling plants named FastPlan. With it you can outline each of the recycling machines and calculate the volume of material to be recycled and the associated costs.

Humara needed help to:

• Automate the collection of information from external databases that feed your volume and cost calculation algorithms
• Automate the generation of plant documentation
• Help the user in the designs

Value provided by Taniwa

Taniwa developed:

• A system to perform ETL (Extract, Transform, Load) to automate the collection of information from external databases
• Deployed a cluster architecture to be able to execute AI processes
• Created an automatic documentation generation system based on LLM (Large Language Model)
• Created a chatbot using LLM based on RAG (Retrieval Augmented Generation)

We have deployed the necessary infrastructure to be able to execute computing-demanding AI processes.

Technologies

• Prefect for ETLs
• Cluster Ray to scale AI and python applications
• OpenAI to use LLMs
• Lang Chains to use RAGs