Not at all! I appreciate the kind words. Thank you!
2,573 karma · joined March 1, 2020
The intended audience was my team and fellow practitioners; assuming some understanding of the jargon allowed me to skip the basics and write more concisely.
The dataset contains the complete post history of over 4M users (81% of all registered accounts), totaling 235M posts. We also make available social data covering follow, comment, repost, and quote interactions.
Since Bluesky allows users to create and bookmark feed generators (i.e., content recommendation algorithms), we also release the full output of several popular algorithms available on the platform, along with their timestamped “like” interactions and time of bookmarking.
This dataset allows unprecedented analysis of online behavior and human-machine engagement patterns. Notably, it provides ground-truth data for studying the effects of content exposure and self-selection, and performing content virality and diffusion analysis.
it's originally "Ready to ~~delve~~ dive in?" but something got lost in translation
On the other hand, directly using an off-the-shelf model, even the best ones, may not meet your performance requirements.
That’s where fine-tuning an open LLM is necessary.
How do you measure the impact of writing a document (vs coding a feature)? What's the right balance between both?
The most impressive candidates can demo something they've deployed. This demonstrates the ability to apply what you've learned AND learn the rest of the stuff as needed (e.g., spin up EC2, build basic front end, maintenance, etc.). The latter is more important and gives confidence that you take ownership and likely can do the same (i.e., end-to-end with results) in the company.
On the second point: having meaning in work differs from person to person. Some people enjoy pushing the envelope (research), others enjoy staying hands-on (individual contributor, IC), while some big picture thinkers enjoy coordinating (PMs). It's helpful to think back on what gave you the greatest satisfaction and work with your leaders to be a role that lets you perform your best.
I find it most fulfilling to build data products that are useful, relative to generating tons of revenue. I would be equally happy as an IC doing the hands-on or leading a small team and mentoring them to become more effective.
With regard to presentation and speaking, I had great leaders around me who provided candid feedback the presentation was off-point, too long, or had to be re-ordered (e.g., when presenting to non-tech folks, results before methodology). It was also useful to "rehearse" any conference presentations at meet-ups and get feedback.
Feedback to improve written communication is harder to come by. Few people feedback on your design documents or reports. For this, I find blogging and seeking feedback from writers whose writing I enjoy to help the most.
It's not difficult, you just need to know that creating an artifact is not the end of your job--it needs to be used to benefit customers and generate value. Remember to measure the impact after your feature/system is launched.
I've seen many interview candidates struggle with this (missing step) too. Those that get past it were hungry enough to want it, as well as followed up with proper validation and measurement.