Data Scientist (Experimentation) at OpenAI. ex-VP of Data @Statsig. I used to be a real scientist.

Seattle, WA
Joined October 2009
Timothy Chan retweeted
Today brings both wonderful but also bittersweet news – @OpenAI has announced its intent to acquire @Statsig. @vijayeraji will become the CTO of OpenAI applications, reuniting with @fidjissimo. 1/
15
13
3
454
98,904
I'm excited to be presenting at ODSC West in San Francisco. Please stop by and say Hi!
We asked @trchan1, Head of Data Science at Statsig, about his upcoming session at #ODSCWest: “A/B testing has really evolved over the years. In my talk, I’ll dive into advanced techniques that tackle challenges like network effects, peeking,...” hubs.li/Q02Rqqnw0
4
413
"in-office culture is stale and boring" Meanwhile at @statsig....
1
4
480
"Data Scientists are not chart makers" - Fun anecdote I shared about pushing back as a Data Scientist against low-impact work.
2
185
Teaming up on this one and hoping to share from our past mistakes and success.
Learn common reasons analytics teams fail and how to make your data efforts count in our upcoming virtual webcast 📹 Featuring Head of Data, @trchan1 & Shachar Meir! Save your spot: us06web.zoom.us/webinar/regi… #datascience #analytics #statsig #abtesting
1
2
196
Definitely going to give this a read
You’re unable to view this Post because this account owner limits who can view their Posts. Learn more
2
188
The one thing we've learned in the last week...
1
3
199
Timothy Chan retweeted
It's officially Day 5 (and final) of Experimentation Week 🖐️ Scaling experimentation is as much a culture challenge as it is a statistical one. Today we share a suite of collaboration features progressively developed into Statsig's core experimentation workflows! 1/5
1
11
63
9,381
Timothy Chan retweeted
Experimentation Week Day 4 - Announcing Interaction Detection 👏 Our customers run 100's of experiments concurrently. We explain how multiple experiments can have an interaction and how you can check for interaction effects on a set of metrics. 1/3
2
8
54
8,676
Timothy Chan retweeted
It's Experimentation Week - Day 2 ✌️ We're announcing one-click Stratified Sampling to make your experiment results more trustworthy and consistent. See how this feature works in Statsig and helps you reduce false positive rates at the same power. 1/3 #datascience #statsig
3
5
1
57
7,068
Timothy Chan retweeted
It's Experimentation Week Day 3 👏 Today, we announce Differential Impact Detection to automatically flag when segments of users respond differently to an experiment - also described as Heterogeneous Effect Detection or Segments of Interest 1/6
2
4
1
60
8,390
Timothy Chan retweeted
Stratified sampling is especially useful for experiments where a tail-end of power users drive a large portion of a metric value, such as in B2B experiments with small sample sizes. Read more: statsig.com/blog/introducing… 2/3 #abtesting #datascience
1
1
165
Timothy Chan retweeted
Day 1 of Experimentation Week is here 📢 We're kicking it off with @trchan1 announcing Meta-analysis! 1/5 #abtesting #experimentation #datascience
5
8
59
7,927
This interview w/Zach Wilson @zachly is deceptively insightful. Many DEs focus on building pipelines but don't ask why. Zach shares ACTUAL tactics and questions DEs should be using to drive business impact. This is how you move to the next level. youtube.com/watch?v=TqI6I1Hn…
1
10
1,040
Timothy Chan retweeted
A Data Scientist, an Engineer, and a Product Manager walk into a restaurant... Featuring: @JiakanWang, @vineethinsea, and @trchan1 #statsigsnapshots
1
2
139
Timothy Chan retweeted
When your leftover billboard helps reinforce @Rippling's new billboards ☝🏻 Featuring Head of Data, @trchan1 , on his recent trip to SF. #statsigsnapshots
2
4
228
Timothy Chan retweeted
Will you be at our A/B Talks in NYC 🍎 Join Statsig's Head of Data, @trchan1, as we host an amazing panel of experts including: - Joel Witten - Head of Data, @recroom - Gaurav Misra - CEO, @getcaptionsapp - Will Ruben - CEO, Stealth Startup Register: eventbrite.com/e/ab-talks-ny…
1
4
141
Public Service Announcement for my fellow SQL practitioners. If you are on BigQuery, Snowflake, Databricks, or DuckDB you can use GROUP BY ALL. The dark days of column counting are over.
4
300