14/03/2026 1:50 am
Topic starter
Notifications
Clear all
Agent Skills Forum
3
Posts
4
Users
0
Reactions
270
Views
0
I'm honestly so pumped right now because I just landed my first proper AI internship at this small fintech startup in Austin starting next month. It's been a long road switching over from pure math and I've spent the last year just grinding on Python and PyTorch and linear algebra but now that it's getting real I'm starting to panic a little about the other side of the job.
I spent all night reading articles on LinkedIn and some blogs about soft skills for data scientists and AI devs but the advice is honestly all over the place. Some people swear that storytelling and data visualization are the only things that matter outside of the code but then I see other folks saying it's all about stakeholder management and being able to explain complex neural networks to people who dont even know what a derivative is. It's kinda confusing because those feel like two totally different skill sets and I dont know where to focus my energy since I only have a few weeks before I start.
- is it about public speaking
- or maybe just project management
- or literal sales skills?
I really want to make a good impression and not just be the math guy in the corner who cant talk to the product team. For those of you actually working in the field what non-technical skills have actually saved your skin or helped you move up because the generic online lists feel like corporate fluff...
3 Answers
11
14/03/2026 2:15 am
I was in your spot a year ago and I am really satisfied with how things turned out after I stopped overthinking the social stuff. Staying organized is what actually kept me safe and my manager had no complaints about my communication. Two quick tips that worked well for me:
- Document the why behind every model tweak, not just the code changes.
- Dont feel pressured to answer everything instantly; accuracy beats speed every time. I have been using Notion Desktop App for Windows to keep my project logs and it works well for staying prepared for meetings. Honestly, if you can show you have a reliable process, the product team will love you. You dont need to be a salesman, you just need to be the person who knows exactly where the data is and why things are happening.
10
14/03/2026 2:15 am
Just found this. Does the startup focus on real-time fraud detection? Be careful with infrastructure costs; I would suggest Weights & Biases Experiment Tracking Service to monitor metrics and justify your GPU spend.
1
14/03/2026 1:50 am
Unfortunately, most online advice is pretty bad and not as good as expected for real work. Had issues before with devs who only talk in equations tho. Not sure but I think these are the real ones:
- Managing expectations so you dont overpromise.
- Translating math into business value.
- Patience. IIRC, most stakeholders just wont get the technical side.
