Predibase

Senior Data Scientist

Amazon Web Services (AWS)

Teaching machine learning pipeline courses.
Building demos and materials for students.

Senior Machine Learning Instructor

Machine Learning Solutions Lab: I work with AWS customers to translate the business problems to ML problems and build the models for them.
 
• Led the project with Sportradar to develop a Soccer Goal Predictor using action recognition models. The solution will help broadcasters with storytelling and fan engagement.
• Led the project with UIUC Football coaches to improve game preparation. Our ML model automated the process and reduced the workflow from hours to under 10 mins, and it provides more insights for the coaches.

Data Scientist

Amazon

A/B Testing (experiment design, causal inference, decision making, metrics deep dive)
Collecting data, generating features and building machine learning models for prediction with customer behavioral data to shorten experimentation time.

Designed success metrics for marketing and business for better and faster decision making
Created data visualization tools and analytics report 

Gave company-wise training and workshops on A/B testing and data-driven decision-making

Created a knowledge sharing internal newsletter on "A/B testing learnings and best practices". Grew the list to 1,000+ subscribers.

Boingo Wireless

Won 1st place of the Boingo Wireless Hackathon by creating a time series model using trigonometric to forecast daily ads capacity. 

Time series analysis
Capacity forecasting

Business Intelligence Data Analyst

SMIF UCI

Student Managed Investment Fund (SMIF) Team

Our team won 1st place in the CFA Orange County portfolio management competition in 2012.

Conducted top-bottom research pertinent to the $33,000 fund’s investment objective, including economic research, industries research, and asset allocation.

Equity Research Analyst

(Opinions are mine, not my employer.)
I'm a data scientist and podcast host. I have worked on machine learning, time-series forecasting, A/B testing, and product analytics. 

I built the first time series forecasting model for advertising at Boingo Wireless; analyzed and help launched over 100 A/B tests at Amazon and have a US patent on experimentation with my colleagues. I built ML models for AWS customers from various industries to accelerate their business and products. I loved it.

I'm from Dalian, a coastal city in China. I lived in LA and Seattle before I moved to SF. I just love the smell of the sea. (...and the smell of coffee!)

If you want to read more career insights I share exclusively with my friends, simply add your email here www.dalianaliu.com and I'd be happy to share them with you. Thanks for connecting with me!

A/B testing -> where intuition meets the unexpected. 

I recently talked to the 4th data scientist at Airbnb on the the unexpected challenges in A/B testing, how to run tests for ML models, and why he built a new A/B testing platform, Eppo. 

Topics we covered:
00:01:26 Getting started in data science at Airbnb
00:03:08 Keys to successful A/B testing
00:06:53 Interpreting A/B test results
00:15:00 A/B testing best practices for ML models
00:53:46 Preparing data scientists for the future
00:59:33 Developing communication skills as a data scientist
01:08:43 Transitioning from individual contributor to manager
01:12:28 The future of experimentation

Learn more from my conversation with Chetan Sharma: 
YouTube: https://lnkd.in/gDAWvuue
Apple: https://lnkd.in/gdp8hk7U
Spotify: https://lnkd.in/g74Gz4xK

What things you want to learn most about A/B testing? Leave a comment and I'll share more later. 

#datascience

Daliana Liu

Open video: A/B testing at Airbnb - Che Sharma - The Data Scientist Show #068 on youtu.be

https://youtu.be/erW_AvYRERo?si=CFpVeiCCx6x8V8H9

Back in 2020, I met Allie K. Miller while we both worked at AWS and she became one of my favorite voices in AI. Not only is she incredibly knowledgeable, having led pivotal AI initiatives at IBM and Amazon, but her approach to AI are truly inspiring. With a following of 1.5 million, it's clear I'm not the only one who thinks so.

Now, she's teaching an "AI for Business Leaders" course, with an impressive lineup of guest speakers - think COO of OpenAI and the Founder of Siri. She already has enrollment from companies like Doordash, Deloitte, Twilio, CVS, Bain Capital, Cisco, Mastercard, and more.

I'm enrolling in the course. If you're eager to elevate your AI leadership skills and join a cohort of like-minded individuals, I'd love for you to be part of this journey with me. Check it out:
https://lnkd.in/g8-3CxVa

Here is a photo of us at a meetup in New York last month, it was great catching up with Allie and I'm looking forward to the course! 

#datascience #ai

Finetune LLM on your own data:

The first problem you’re likely to encounter when fine-tuning an LLM is the “host out of memory” error. It’s more difficult for fine-tuning the 7B parameter Llama-2 model which requires more memory. The good news is that, with an optimized LLM training framework like Ludwig, you can get the host memory overhead back down to a more reasonable host memory even when training on multiple GPUs.

I'd like to share a hands-on workshop tomorrow from DeepLearning.AI that shows you how to tackle this. By the end of this session, you will understand:
• How to fine-tune LLMs like Llama-2-7b on a single GPU
• Techniques like parameter efficient tuning and quantization, and how they can help
• How to train a 7b param model on a single T4 GPU (QLoRA)
• How to deploy tuned models like Llama-2 to production
Continued training with RLHF
• How to use RAG to do question answering with trained LLMs

This session will equip ML engineers to unlock the capabilities of LLMs like Llama-2 on for their own projects. 
Register here (free): https://lnkd.in/eV7JkWQM

You can watch the recording if you register, even if you can't make it.

Tag a friend who might find it useful.

--------
S﻿peakers:
• Piero Molino is the co-founder and CEO of Predibase. He was one of the founding members of Uber AI Labs. Creator of Ludwig.
• Travis Addair is co-founder and CTO of Predibase. In the past, he led Uber’s deep learning training team as part of the Michelangelo machine learning platform.

#machinelearning #llm

Let's meet up in New York!

I've been writing on Linkedin for over three years but haven't met any readers in NYC yet. Let's change that. Come join me this Sat (Sep 2nd) morning, register here: https://lu.ma/1hvbxgu4

I also invited guests from "the data scientist show" and can't wait to talk about data science, career, and life. It'll be a great start for the long weekend, and I'm so excited to meet you!

(And if you live in Los Angeles, I'm organizing one next weekend -> register here: https://lu.ma/igg3pc8r)

Tag a friend who might be interested :)

#datascience #career #newyork

Let's do another meetup!

I met a few long-time podcast listeners and LI followers in New York and it's great to talk about their journeys and the future of data science.

If you live in LA, come join me and Damien Benveniste, PhD this Sunday morning. Register here: https://lu.ma/igg3pc8r

Tag a friend who might be interested!

If you don't want to miss future meetups in your city, subscribe here: www.dalianaliu.com

#datascience

If you want to learn NLP, don't just read papers - learn how to apply it. 👇🏼

The wold’s largest applied NLP community is hosting a FREE virtual Summit (join here: https://lnkd.in/gyCYMkk8) What you'll get: 
🔥 50+ technical sessions aimed at applied learning.
🧠 Deep dives into Open Source, Healthcare & Applications.
🤔 Live Q&As to tackle those burning questions.
🤝 Networking opportunities to share your struggles and triumphs.
🔧 Week two: getting hands-on. 

You should expect intensive training workshops to turn your frustrations into expertise. 

Date: Oct 3-5, 2023
Free registration: https://lnkd.in/gyCYMkk8

Tag a friend who might be interested! Which session are you most excited about?

#datascience #machinelearning #nlp

Gen AI & Machine learning (in the enterprise!) 

It's easy to build a small project using Gen AI, but there are challenges in building production-grade data&ML products. It all comes down to performance, cost, and trust. 

If you want to get in the discussion with Gen AI and ML leaders from Meta, JP Morgan, Visa, AT&T, Blackstone, Stanford and more, apply to the Data Science Salon Conference in Miami: https://lnkd.in/gHaHYRmz

One of the speakers Laura Gabrysiak has also been on "the data scientist show" and I look forward to her talk! 

The is a very limited amount for the complimentary conference pass you can apply here before it expires: https://lnkd.in/gHaHYRmz

Tag a friend who might be interested. What topics are you excited about? Leave a comment.

#genAI #machinelearning

Taplio

Daliana Liu's Linkedin Analytics

Daliana Liu

Who is engaging with Daliana

Daliana Liu's Best Posts (last 30 days)

Want to drive more opportunities from LinkedIn?

Taplio

.css-1jgreyo{background-image:linear-gradient(284deg, rgba(39,43,72,1) 0%, rgba(33,47,183,1) 100%);color:transparent;-webkit-background-clip:text;background-clip:text;}Daliana Liu's Linkedin Analytics

Daliana Liu

Who is engaging with Daliana

Daliana Liu's Best Posts .css-qqfgvy{font-size:var(--chakra-fontSizes-md);}(last 30 days)

Want to drive more opportunities from LinkedIn?

Daliana Liu's Linkedin Analytics

Daliana Liu's Best Posts (last 30 days)