Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
企業開發文化: MLOps 面向
Search
LINE Developers Taiwan
PRO
January 16, 2024
0
3.3k
企業開發文化: MLOps 面向
Speaker: Rei Huang
Event: AWS Educate 企業參訪
LINE Developers Taiwan
PRO
January 16, 2024
Tweet
Share
More Decks by LINE Developers Taiwan
See All by LINE Developers Taiwan
透過 Cloud Functions, Firebasem Gemini Pro 打造具有記憶的 Open Data LINE Bot
line_developers_tw
PRO
0
10
就讓AI幫你帶LINE好運
line_developers_tw
PRO
1
32
沒想過的前端錯誤處理可能比你有做的還多
line_developers_tw
PRO
0
51k
LINE 台灣實習計畫 2024 - TECH FRESH 介紹
line_developers_tw
PRO
0
510
LINE 台灣實習計畫 - 相關福利與面試流程
line_developers_tw
PRO
0
520
Navigating My Internship: A 5W1H Perspective
line_developers_tw
PRO
0
23
1001 號 到 333 號
line_developers_tw
PRO
0
15
關於我連線三年半的故事
line_developers_tw
PRO
0
870
LINE 電商部門實習分享
line_developers_tw
PRO
0
48
Featured
See All Featured
Testing 201, or: Great Expectations
jmmastey
30
6.4k
Thoughts on Productivity
jonyablonski
60
3.9k
Building Better People: How to give real-time feedback that sticks.
wjessup
356
18k
Into the Great Unknown - MozCon
thekraken
15
1k
How GitHub (no longer) Works
holman
305
140k
We Have a Design System, Now What?
morganepeng
43
6.8k
GraphQLの誤解/rethinking-graphql
sonatard
56
9.3k
Principles of Awesome APIs and How to Build Them.
keavy
121
16k
Building an army of robots
kneath
300
41k
Creatively Recalculating Your Daily Design Routine
revolveconf
211
11k
[RailsConf 2023] Rails as a piece of cake
palkan
28
4k
The Illustrated Children's Guide to Kubernetes
chrisshort
32
47k
Transcript
None
企業開發⽂化 MLOps ⾯向
01 02 03 04 What is MLOps? AS IS &
TO BE What should do first? How to do? Team Strategies Self Introduction 05 Q & A CONTENT MLOps: Machine Learning Operations
01 Self Introduction
Rei Huang Machine Learning Engineer Education • Bachelor in Mathematics
@ CCU • Master in Mathematics @ NTHU Work Experience • E.Sun Commercial Bank (2016 ~ 2022) • LINE Taiwan EC (2022 ~ Now)
MLOps: Machine Learning Operations 02 What is MLOps
圖片來源: D.Sculley et. al. NIPS 2015 Hidden Technical Debt in
Machine Learning Systems 2100 特點項⽬⽂字 特點項⽬ The surrouding Infra of ML code 1234 特點項⽬⽂字 特點項⽬ 999 特點項⽬⽂字 特點項⽬
Actual Task Flow Discuss Business Demand Assess demand by data
Create Issue or ticket Start coding Test PR Deployment Monitoring Design
Actual Task Flow Discuss Business Demand Assess demand by data
Create Issue or ticket Start coding Test PR Deployment Monitoring Design
廣度? How to complete such a big deal? 圖片來源:Yahoo 新聞:數位人才有哪些能力?數位領域的T型人才驅動企業成功數位轉型
API Database ML Algorithm Data Analysis
03 AS IS & TO BE
⼯作上會這麼做 你現在可能這樣做 Notebook 寫到底 一個 script 寫完所有邏輯 使用 ensemble 建模型把
performance 最佳化 有很多時間嘗試不同的 ML 演算 法,甚至可以 tune hyper parameters 有結構化的 production code 完整的建模流程與服務架構 需要考量訓練與預測的運算效率 穩定的 model serving 不同應用情境的 feature engineering 圖片來源:onlyGFX.com
Real Work Style 開發前與大家同步執行方式 or design 程式碼是需要被 review 的 (Pull
Request) 先講解 PR 目的 Review PR 後直接與原作者溝通 頻繁且快速 sync Documentation (Coding as document) 共同制定 team best practice 分享、回饋、調整
What should do first? How to do? 04 Team Strategies
時間有限 but 願望無窮 To do or not to do, that
is a question 新的 business 需求 新的 feature 加入訓練資料 Clean Code Unit Test 想要測 試更多情境 Machine learning Metric & business evaluation CI / CD 更完整 Log format 想要修改 的更漂亮 看到新的 model 想要 套用 多組 model 做交叉比對 對特定資料或者情 境做資料分析 圖片來源: ⾃由時報 莎⼠比亞的10個⼩祕密 你知道幾個呢?
實務上的抉擇基準 P0: 最緊急 P1: 如果沒有緊急事項 就先做 P2: 不急,有空再做 Feature: 新功能
(P0 or P1) Bug: 現在不修就完了 (P0) Known issue (P1 or P2) 解決方式 短解:僅解決現有問題 長解:擴充功能、大規模 refactor or enhancement 問題種類 緊急程度
結論 我們以為的 ML、AI 工程師不僅是做 ML、AI 真實的企業開發與學生時代的 research、比 hackathon、打 kaggle 有不少差異,需要很多團隊合作以及
為商業場景做設計 策略的重要性、團隊的核心價值 Work Smart / Work Hard
None