kn81198

Location: Seattle, WA

Remote: Yes

Willing to relocate: Yes (Washington, California)

Technologies: Python, PySpark, Go, SQL, C++, CUDA, PyTorch, scikit-learn, XGBoost/LightGBM, Kafka, Airflow, Vertex AI, Kubernetes, Docker, GCP, AWS

Resume: https://knhash.in/resume

Email: [email protected]

ML Engineer, 6 years shipping real-time recsys and ML platform systems at 100M+ DAU, sub-50ms p99. MS CS Georgia Tech (May 2026), HPC specialization, analog CiM research.

Looking for founding engineer, ML Engineer roles in recsys, ranking, ML platform, ML systems, or HPC.

kn81198

2 か月前議論

True, but the extra smarts were more of an inconvenience rather than a big issue. The standalone part was useful - we needed to keep the number of parts (therefore the points of failure) low. For instance Unity made it super easy to build simple video loop and not having to worry too much about the FPV settings of a display and the built in battery worked to keep it portable.

kn81198

3 か月前議論

What do you think of making it a spaced repetition-esque system where you treat hitting the snooze button repeatedly as a back off signal?

kn81198

6 か月前議論

https://knhash.in My home on the internet. It also has cat picture. And is about comedy. Sometimes.

kn81198

2 年前議論

Vipassana Meditation formalizes this to a large extent and is extremely approachable. I would highly recommend anyone looking at guided meditation to give this a try:

https://www.dhamma.org/en-US/index

投稿

The Remarkably Human Samsung Unsubscribe Flow

You want to build a Recommender System

Rage-Inducing Problems in Tech

US now spending more on data center than on public transportation

How to reduce the cost of generating an Image

Being a line cook without talking to AI

Show HN: I Built a Museum Exhibit

Starting from Scratch

Intuitive Understanding of PCA

Nifty and Its Derivatives; and Arbitrage

Expect Subvertations

コメント