What I'm Thinking About
Data as a living system: Every dataset tells a story, but most people stop at chapter one. I've learned that the real insights come from understanding how data evolves - whether it's tracking patient outcomes over years or watching market patterns shift in milliseconds.
Teaching is debugging for humans: Being a TA for 40+ students taught me that education is just debugging at scale. The best part? Unlike software, students debug themselves once you give them the right tools.
Pattern recognition everywhere: From optimizing database queries to building competitive Pokémon teams, it's all about finding patterns others miss. My approach? Look for the connections nobody else is making - that's where the magic happens.
"Every system can be optimized. Every recipe can be improved. Every Pokémon team can be beaten. The fun is in trying."
— My engineering philosophy
Problems I Love Solving
"This query takes forever to run"
My best: 48 hours → 15 minutes for genomic analysis. The secret? Usually it's not optimization, it's completely rethinking the approach.
"We have the data but can't make sense of it"
1.3TB of clinical trial data taught me that more data ≠ more insights. The art is knowing which 0.1% matters.
"Can ML actually help here?"
Sometimes the answer is no, and that's okay. Knowing when NOT to use ML is as important as knowing when to use it.
Currently Exploring
🏥 Healthcare Data Standards
Building bridges between different medical systems. Because your health data shouldn't need a passport to travel between hospitals.
⚡ Real-time Analytics at Scale
Exploring how to make big data feel small. Current experiment: processing 100k events/second on a budget smaller than my coffee expenses.
🔮 Predictive Maintenance
Teaching machines to predict their own failures. It's like fortune telling, but with more math and fewer crystal balls.
🎯 Decision Intelligence
Combining human intuition with machine precision. Because sometimes the best algorithm is knowing when not to use one.
Things I've Built
See all projects →Text-Conditional Chest X-ray Generation Using Latent Diffusion
Resource-efficient medical image generation with 148.84M parameters
Multi-omics Data Processing Pipeline
192x performance improvement for genomic analysis at Harvard Medical School
Real-Time Stock Market Data Processing Pipeline
Sub-second latency pipeline processing 10,000+ messages per second
Random Facts About Me
- • I once debugged a data pipeline at 3 AM by realizing the issue was daylight saving time
- • My Pokémon team has better type coverage than most Fortune 500 data architectures
- • I have a year of culinary school under my belt (left when I realized vegetarians don't make great butchers)
Thanks for making it this far. Hope you found something interesting. ✌️