---
title: "Baku Kim (@probaku1234) | Hashnode"
description: "AI Engineer | Ex-Backend Developer | Tech Enthusiast. Read the latest articles by Baku Kim on Hashnode."
url: "https://hashnode.com/@probaku1234"
---

@probaku1234

# Baku Kim

@probaku1234 · Joined February 2025

AI Engineer | Ex-Backend Developer | Tech Enthusiast

### [Mimicking Jev with vLLM: A System One-Style Decision Endpoint from Next-Token Probabilities](https://bakus-tech-blog.hashnode.dev/mimicking-jev-with-vllm-a-system-one-style-decision-endpoint-from-next-token-probabilities?utm_source=hashnode&utm_medium=feed)

Build a vLLM plugin that adds /v1/systemone and scores choice, score, and yes/no questions using a causal LLM’s next-token log-probabilities.

20 min read · [bakus-tech-blog.hashnode.dev](https://bakus-tech-blog.hashnode.dev/mimicking-jev-with-vllm-a-system-one-style-decision-endpoint-from-next-token-probabilities?utm_source=hashnode&utm_medium=feed)

[Nov 14](https://hashnode.com/posts/mimicking-jev-with-vllm-a-system-one-style-decision-endpoint-from-next-token-probabilities/6af7fff546697869524c3a9f)

### [Understanding Unpin and Pinning with tokio::select!](https://bakus-tech-blog.hashnode.dev/understanding-unpin-and-pinning-with-tokio-select?utm_source=hashnode&utm_medium=feed)

Learn why tokio::select! requires Unpin futures and how to fix compiler errors using tokio::pin! vs Box::pin for stack and heap pinning in Rust.

6 min read · [bakus-tech-blog.hashnode.dev](https://bakus-tech-blog.hashnode.dev/understanding-unpin-and-pinning-with-tokio-select?utm_source=hashnode&utm_medium=feed)

[Jul 26](https://hashnode.com/posts/understanding-unpin-and-pinning-with-tokio-select/6a658964ee30bd18d9bd433f)

### [Part 7: Backward Pass](https://bakus-tech-blog.hashnode.dev/part-7-backward-pass?utm_source=hashnode&utm_medium=feed)

Up to this point, We have built a 'Forward Pass' that makes predictions, and a 'Loss Function' that measures exactly how far off those predictions are. In order to enable our model to learn patterns f

11 min read · [bakus-tech-blog.hashnode.dev](https://bakus-tech-blog.hashnode.dev/part-7-backward-pass?utm_source=hashnode&utm_medium=feed)

[Feb 24](https://hashnode.com/posts/part-7-backward-pass/699d7822003fddb095d5cb16)

### [Part 6: Data Scaling – Normalization vs. Standardization](https://bakus-tech-blog.hashnode.dev/part-6-data-scaling-normalization-vs-standardization?utm_source=hashnode&utm_medium=feed)

In our previous part, we witnessed a "numerical explosion." Because our target formula involves x³, the raw values reached into the thousands, causing our loss to skyrocket into the millions. To build

8 min read · [bakus-tech-blog.hashnode.dev](https://bakus-tech-blog.hashnode.dev/part-6-data-scaling-normalization-vs-standardization?utm_source=hashnode&utm_medium=feed)

[Feb 24](https://hashnode.com/posts/part-6-data-scaling-normalization-vs-standardization/699d777b76cf0888f4a43c4b)

### [Part5: Forward Pass](https://bakus-tech-blog.hashnode.dev/part5-forward-pass?utm_source=hashnode&utm_medium=feed)

With our dataset prepared and our model's architecture ready, we are at starting point of deep learning. The first part is the forward pass. In this part, we will implement the entire forward pass fro

10 min read · [bakus-tech-blog.hashnode.dev](https://bakus-tech-blog.hashnode.dev/part5-forward-pass?utm_source=hashnode&utm_medium=feed)

[Feb 24](https://hashnode.com/posts/part5-forward-pass/699d770c76cf0888f4a4152a)
