How can I run Spark NLP on an Orin GPU over http and use Postgres?

[KnowledgeGarden]

I have a Jetson Orin 64gb to which I can easily send text bodies for GPU work over http, though there may be a way to install some part of SparkNLP on that tiny box. My preference is to use Postgres rather than, say, neo4j. I'd appreciate any insights on this idea. It's large enough for a fairly robust OSS LLM. Thanks in advance.