1. What can I do
2. What I’m probably suitable
3. What will I do

It’s about time to find a job, here I record what I’m probably suitable for, what can I do, what will I do.

What can I do

These skills I can only call them “basic”, proficient is on way to legend.

• Python, C, C++, git, CMake, yaml
• STM32 series embedded system dev
• CI/CD test
• Generic software programming (if for interviews, can kill leetcode mid)
• very basic ML

What I’m probably suitable

• Embedded system engineer (not sure yet)
• Generic software/Algo engineer (almost need strong programming skill ig)
• Big data dev (need DB + basic ML)
• ML engineer (need ML + basic DB)
• Backend dev (need network knowledge)

What will I do

• SQL, Spark
• ML basic
• TCP/UDP basic
• Operation system
• Embedded linux

1. Baidu
   1. 1st interview:
   2. 2nd interview:

This page records how I find a job. I write them down at the time I still remember them. Everything you write on the resume should be things you know.

Baidu

• Job: ML/DM/NLP engineering
• Timeline: Ended in 1 week, resume “shared”
• Comment: They called me to ask if I’m willing to try this job. Interviews proceeding in RuLiu app.
• Records:

1st interview:

Self introduction, 3 algo.

Construct a binary tree from it’s preorder and inorder sequences, and output the breadth order of tree nodes.

Mid, find the root in preorder and inorder array, recursively construct the BT, then solve with BFS, need some lines of code.

Coin change II. Given several kinds of coins, output the number of methods to reach the amount.

Mid, basic dp, like fibonacci. For 5, [1,2,5], we can have a dp array [1,2,2,3,4], 4 methods to reach amout 5. I’m a bit lazy to write a math equation on this web page, but it’s a fibonacci like dp, observe and solve.

House robber. Given an array of houses, rob the most money. Rob adjacent house will trigger alarm.

Mid, we rob the current house based on the former 2 houses, because we can’t rob adjacent houses. So the current dp, with index i, is max(dp[i-1], dp[i-2] + house[i-1]), house’s index is based on your counting, it’s the former of the former house.

2nd interview:

Self introduction, ml knowledge, 1 algo.

About my project

Well it’s not a pretty funny project, no such high skills like other cs masters, just a bit related to ml.

What’s Adagrad? What’s weight?

Adagrad is evolved from SGD, added $s_t = s_{t-1} + g_t * g_t$, allowing parameters to update by different time. A con is, $s_t$ grows larger along with time, making $\eta$ smaller and smaller, finally learns nothing after many iterations.

Weight is a kind of parameter in nn, indicating the information conducts from one neural to another. It will be updated along with iterations.

• Thoughts: of course I didn’t make them all in the interview, my coding skill would pass, but not my emotions and basics of nn. I was too nervous to answer these questions. Guess it’s the reason I fall from the sky.

Update TIC-80’s api of pkpy from v1.x to v2.0.1.TIC80 Pkpy a python interpreter for game scripting.Pkpy

CPP test

28 Oct 2024 in Coder / Cpp on Coder, Cpp

This is a test post

1. type some code

type some code

#include<iostream>
int main()
{
std::cout<<"say hi"<<std::endl;
}