Instagram story viewer> @maggieindata> Posts

Maggie | Data Science & AI

@maggieindata

88.1K

followers

550

following

Level up your tech career with AI
👩🏻‍💻 ex-tech data scientist -> building in public @tryswpt
📍 toronto / LA
📧 partnership: [email protected]

POSTS STORIES REELS TAGGED

Comment CHAT and I’ll send you the link to try codex for yourself! #ChatGPT_Partner

It is genuinely mindblowing how the future of AI assisted data analysis is already here with @openaidevs codex tool!

An isochrone map like this used to take a full day to build when I worked at the planning office with the city. It would often include using proprietary GIS softwares, wrangling with GTFS data formats, coordinate transforms and projections, and running feeds through multiple tools.

I just generated this with codex in 8 minutes.

What Codex actually did here: pulled GTFS data -> mapped real walking catchments -> identified the transit deserts -> output files in HTML + PNG + GeoJSON format that are actually commonly used in geospatial data science and in the context of urban planning

#DataAnalysis #WomenInTech #CareerinData #DataScience by @maggieindata

169

42

5 months ago

Which one hit too close to home for you? The unspoken truth about how people feel working in tech but are not really talked about. I’ve experienced way too many times when company’s priorities suddenly shift and your project is being put on the back burner

#workingintech #instagramvsreality #datascience #breakintotech #programmerjokes by @maggieindata

15k

173

a year ago

Normalize outgrowing.
Normalize wanting more.
Normalize choosing yourself without guilt ✨

Staying where you’ve already outgrown yourself?
That’s the real risk.

If this hit… you already know it’s time 👀
Follow for real, no-BS career advice 🚀

#careergrowth #careerchange #careeradvice #careerjourney #growthmindset by @maggieindata

7k

64

6 months ago

3 data science portfolio project ideas without machine learning:

1️⃣ practice an A/B test with simulated outcomes
2️⃣ map patterns in NYC 311 requests
3️⃣ build a map of available bikes in Toronto.

Each one gives you a different problem to work through: comparing groups, cleaning messy records, or turning API data into something useful.

Then explain your decisions. Why did you exclude those rows? What does your result actually tell us? What would you need to check before someone acts on it?

Keep simulated outcomes clearly labeled and be specific about what the data cannot answer.

💌 Save these three ideas for your next portfolio project. by @maggieindata

57

2

10 hours ago

Somebody had to say it by @maggieindata

166

4

2 days ago

An AI saying “I might be wrong” doesn’t automatically make it trustworthy.

In one example described by Anthropic, a model learned that adding more warnings could earn it points—without making its answers more honest. That’s reward hacking: optimizing for the score instead of what we actually care about.

Uncertainty can be useful. But it isn’t proof. Check the example. Open the sources. Test the code. Trust what holds up.

Save this for your next AI rabbit hole. 👀

#AI #AIliteracy #ArtificialIntelligence #MachineLearning #RewardHacking by @maggieindata

3

1

3 days ago

Your biggest career opportunities won’t happen behind a screen.

I went to a conference or in-person event at least once every month, and I deeply believe this is the reason why I was able to 7x my business revenue this year.

I’m a huge believer in showing up, getting into rooms where I’m the youngest person or the one with the least amount of knowledge, and actually talking to people building what’s next.

The next room ➡️ @microsoft Ignite

I’ll be in San Francisco November 17-20. LET ME KNOW if you’ll be there also!

🔥Comment ‘IGNITE’ and I’ll send you the link! (online registration is free)

#MicrosoftPartner #MSIgnite #CareerinAI #WomeninTech #DataAnalytics by @maggieindata

3

7

6 days ago

He built a fully automated process for content creation from research to scripting to editing and publishing with AI agents 👀 would you try editing with AI?

#careerintech #aiedit #datascience #entreprenuer by @maggieindata

3

1

6 days ago

Would you return to data science for $500K? Maggie said no 👀
ㅤ
I sat down with Maggie Ma to talk about why she left tech after working so hard to become a Data Scientist.
ㅤ
Her first year after leaving was difficult, but today, she earns more than she did working full-time in tech.
ㅤ
Her advice? Start building while you still have something stable.
ㅤ
Comment “500K” and I’ll DM you the full conversation 👇
ㅤ
Want to be a guest on my podcast @outofofficewithsundas?
Let’s chat: sundaskhalid.com/podcast
ㅤ
#datascience #careergrowth #contentcreator #careerchange by @maggieindata

348

27

8 days ago

Leaving my data scientist job to create full time was my leap. Yours might be applying for a new role, starting a side project, or finally trying that AI tool you’ve been curious about.

I share data science interview prep, the latest AI, and the messy middle of figuring out life and work in my 20s. I’m learning to try things before I feel ready, let some of them fail, and enjoy where I am along the way.

A reminder for me, too 🤍 by @maggieindata

85

3

9 days ago

I built an AI automation that can turn any videos into highlight clips with @chatgpt 🤯😮‍💨

#ChatGPT_Partner by @maggieindata

27

0

9 days ago

You already told your AI assistant that things changed. So why is it still using the old information? 👀

The answer has to do with how AI memory works.

Imagine your team migrates from PostgreSQL to BigQuery, but your coding assistant keeps writing SQL for the old database. It remembered a fact that’s no longer current.

Building reliable AI agents means managing how memories are updated, retrieved, and summarized—while preserving useful history.

When testing an AI agent’s memory, ask:
→ Does it use updated information?
→ Can it still answer questions about the past?
→ Do its summaries preserve key details?

Has an AI assistant ever used information you’d already corrected?

#AIMemory #AIAgents #GenerativeAI #LLM #ContextEngineering by @maggieindata

3

1

10 days ago