What I Accomplished This Week
This was one of those productive weeks where everything just clicked. Here’s what got done:
Jekyll & Site Infrastructure
- ✅ Fixed that annoying build bug that only appeared in Vercel
- ✅ Finished the homage clone styling for Sumedh’s website ahead of the 31st deadline
- ✅ Wrote and posted the August 31st Homage post
Data & Analysis
- ✅ Improved calendar scrapers & importers
- ✅ Updated When Do I Write Code with data from my own GitHub profile
- ✅ Posted my update as a blog entry on this site
- ✅ Analyzed golf statistics using Sumedh’s posts to explain neural network performance
MCP Server Development
The real focus this week was on MCP (Model Context Protocol) servers. I:
- ✅ Created an MCP for my various important filesystems
- ✅ Built an MCP server that auto-identifies appropriate schemas out of HTML code
- ✅ Read OpenAI’s MCP Server specs and implemented one
- ✅ Parsed this comprehensive list of JSON struct definitions
- ✅ Explored TypeScript data tools
Schema.org Integration
- ✅ Followed best practices for integrating schema.org and JSON data
- ✅ Got the schema auto-identification working smoothly
What’s Next
Looking ahead, I’m focusing on:
- Data Aggregation MCP: Building an MCP server that scrapes emails, docs, messages, and websites to create TODO lists (using Bright Data)
- ML Commons Contribution: Getting more involved with the AI Risk & Reliability working group
- BrightData Integration: Scaling up my SingleSiteScraper with BrightData
It’s been a productive week. Now to keep the momentum going!