Timeline

Week	Description of Task	Status
W1	Decide on project and submit initial proposal	Completed
W1	Investigate feasibility of retrieving Microsoft Teams messages as initial dataset	Completed
W1	Write initial script to collect dataset of all security alerts from Teams. (Source code here)	Completed
W1 / W2	Perform initial research on ML, LLM, and agentic RAG and construct initial library of resources	Completed
W2	Collect dataset of all security alerts from Teams	Completed
W2	Collect dataset of all internal wiki pages	Completed
W2	Write initial script to collect dataset of all pages and policies from the organization's public security site. (Source code here)	Completed
W2	Collect dataset of all security web pages, incluing policies	Completed
W3	Research proper data formats for training AI models	In Progress
W3	Build a local LLM test environment for deploying new models	Completed
W3	Research existing LLM models as a potential base model	Completed
W4	Build formal outline of the project, specific data sets in use, and testing methodology for verification	Completed
W4	Construct a set of scripts to convert raw datasets to proper format for ingestion	Completed
W4	Convert datasets to proper format for ingestion	Completed
W5	Construct a formal technical architecture diagram for the project	Completed
W5	Test potential LLMs for usability and accuracy of responses	In Progress
W5	Test LLMs against existing alerts and synthetic alerts to establish a baseline of results	In Progress
W5	Begin training LLM with alert data based on prior research	Completed
W6	Continue to train and tune LLM with existing security data	In Progress
W6	Build API for interacting with the model	Not Started
W7	Build Teams webhook pipeline for auto-populating data	Not Started
W7	Build mechanism for security operations team to provide feedback on responses	Not Started
W7	Test model with synthetic alerts to assess responses	Not Started
W8	Implement “production” Teams alert pipeline	Not Started
W8	Collect responses from Security Operations team members	Not Started
W8	Develop plan for evaluating Security Operations responses	Not Started
W9	Measure responses from Security Operations team to evaluate accuracy	Not Started
W9	Build statistics for final results	Not Started
W9	Use results to further train model	Not Started
W9	Identify further areas of improvement to the solution based on responses	Not Started
W9	Explore solution limitations, future work, etc.	Not Started
W10	Write Final Project Report	Not Started
W11	Write Final Project Report	Not Started