https://www.linkedin.com/in/yzouyang/
https://github.com/KunojiLym/databricks-linkedin-analytics
https://www.yzouyang.com/category/blog-series/build-your-own-linkedin-analytics/
Roast My Stack - A Production-Style Personal LinkedIn Analytics Pipeline.pptx
DataBrick - cloud-based platform for data analytics and artificial intelligence. https://login.databricks.com/signup
How to get data?
ETL in medallion terms E → Extract - bronze T → Transformation - silver L → Load - gold
To overcome time out issue —> consider retries and back off method especially for LLM calls and APIs - to prevent a huge bill shock
One Big Table approach over relationship database (e.g neo4j) —> more efficient in terms of query speed and simplify data consumption
https://www.linkedin.com/in/eeonelim/
https://github.com/EeOneLim/care-compass
https://care-compass-iota.vercel.app/
Navigating Care for Singapore's Caregivers.pptx.pdf
RAG over LLM —> if you have single source of truth, RAG is preferred increase reliability
If you need the LLM to do a web search —> use a web search tool for better
Consider n8n dynamic model selector to choose the best model for your usecase also N8N reduces usage of LLM N8N can switch between using LLM websearch and RAG
Consider Postgres https://www.postgresql.org/ for RAG
Consider other LLM for this usecase, GPT 5.4 mini or other open-weight model for cost efficiency
Other tools mentioned: Eval tool in Claude platform
Kapso for whatsapp business workflow https://kapso.ai/
https://www.linkedin.com/in/kaleb-nim/
https://github.com/Kaleb-Nim/nim-kaleb
ASR - Automatic speech recognition
For budget option: Oracle cloud - free tier forever
24GB ram 4cpu
10 TB Egress
Whisper TTS for speech input
always consider guardrails for ai applications simplest way to build guardrails is via personna.md