Roadmap
Below are the projects currently in the roadmap for this course. They are ranked by current priority, although ranking can change based on on-going people requests and/or new AI developments. If you have requests which are not below, just email me and I will add it. Also, if you are particularly interested in one of the items below, email me to let me know so I can update the ranking accordingly.
The ranking depends on a combination of (a) more requests from people -> higher ranking, (b) in line with the spirit of the course, the more applied/related to actually building your own product -> higher ranking, and (c) easier/faster -> higher ranking.
Also, whenever there is some new AI breakthrough, that will go straight to the top of the roadmap and lessons related to it will be added to the course.
| Item | Status | Notes | ||
| Add to the course a Chatgpt App that can answer all sorts of Product Data Science questions. The model leverages personal notes, insider knowledge, all the documents collected over the years, case studies, etc. Eventually, it should be able to answer any product data science case study as well as specific questions about data science at top tech companies | Completed - DataScienceGPT | Looking really cool | ||
| Add to the course a chain that allows to do English -> SQL -> query results, using the coding course queries to validate it/test it. | On-going. Currently working on it. | Lots of requests for this. The issue is that it is not a solved problem if the query isn't trivial. It is easy to implement it for simple stuff/toy examples (group by, top X, etc.), but data scientists don't really write such simple SQL queries in real life. | ||
| Add to the course an add-on option that allows people to buy X USD amount of OpenAI credits along with the course | Likely starting on this after the chain English -> SQL -> query results is done | Already received several requests for this item. This essentially allows people to buy Open AI credits with their employee training budget. Going through the course is very cheap, but experimenting with gpt4 when building an app/business can get very expensive. For this reason this would be very beneficial to people. However, implementation is not trivial. Leveraging this project looks promising though | ||
| Add Python Shiny Code | Either before or after OpenAI credits | Streamlit growth has been so rapid that Python Shiny has been a bit overshadowed as a tool for DS to put up apps. However, there have been interesting developments there too. This is not too hard and the priority rank will likely depend on the number of requests | ||
| Add R code to the Langchain lessons | Will likely do it together with Python Shiny | Implementation is easy via reticulate calling Python. Not too sure about the actual benefits over a system call since reticulate is likely more memory intensive than a system call (at least that's my understanding so far). Since it is easy, it might go up in the priority rank if I get more requests for this |