top of page
Writer's pictureAmar Harolikar

Maybe leave programming to AI. Coding by GPTs: A Data Scientist's Perspective

Updated: Oct 24, 2024

The title of the post is a quote from Jensen Huang, CEO of NVIDIA.ย Couldn't agree more. And a sentiment shared by many fellow analysts.





With over two decades coding and executing data science and analytics projects, GPTs have significantly increased my efficiency. Particularly in code generation, allowing me to focus more on output quality and deployments.


๐—š๐—ฃ๐—ง ๐— ๐—”๐—š๐—œ๐—–ย 

๐—–๐—ผ๐—ฑ๐—ถ๐—ป๐—ด: Best done by AI. Just can't beat the quality and the speed. Ginormous time savings



๐—ฃ๐—ฅ๐—”๐—–๐—ง๐—œ๐—–๐—”๐—Ÿ ๐—–๐—ข๐—ก๐—ฆ๐—œ๐——๐—˜๐—ฅ๐—”๐—ง๐—œ๐—ข๐—ก๐—ฆ

โ–ช ๐—œ๐—บ๐—ฝ๐—ฎ๐—ฐ๐˜: A beautiful code by itself doesn't create any impact. It's part of a bigger pipeline.


โ–ช ๐—š๐—ฟ๐—ถ๐—ป๐—ฑ ๐—ฆ๐˜๐—ถ๐—น๐—น ๐—˜๐˜…๐—ถ๐˜€๐˜๐˜€: e.g. API stuff was new to me (FastAPI, GET/POST, transfers). Understanding took time... and then coding was a breeze as I knew exactly what to ask of GPTs


โ–ช ๐—œ๐˜'๐˜€ ๐—ป๐—ผ๐˜ ๐—ถ๐—ป๐˜€๐˜๐—ฎ๐—ป๐˜ ๐—ฐ๐—ผ๐—ณ๐—ณ๐—ฒ๐—ฒ: We still need to iterate, check outputs, modify code, debug... until it works. No change there. GPTs do make it faster.


โ–ช ๐——๐—ผ๐—บ๐—ฎ๐—ถ๐—ป ๐—˜๐˜…๐—ฝ๐—ฒ๐—ฟ๐˜๐—ถ๐˜€๐—ฒ ๐— ๐—ฎ๐˜๐˜๐—ฒ๐—ฟ๐˜€: e.g. never worked on healthcare analytics. No amount of GPT code would make me an expert quickly


โ–ช ๐—Ÿ๐—ฎ๐—ป๐—ด๐˜‚๐—ฎ๐—ด๐—ฒ ๐˜‚๐—ป๐—ฑ๐—ฒ๐—ฟ๐˜€๐˜๐—ฎ๐—ป๐—ฑ๐—ถ๐—ป๐—ด ๐—บ๐—ฎ๐˜๐˜๐—ฒ๐—ฟ๐˜€: Things go very fast with Python. But, React-Node.js web app? I realized quickly that just dumping codes that I don't understand doesn't work well. Though GPTs do speed up learning.


โ–ช ๐—š๐—ฃ๐—ง๐˜€ ๐—บ๐—ฎ๐—ธ๐—ฒ ๐—บ๐—ถ๐˜€๐˜๐—ฎ๐—ธ๐—ฒ๐˜€ and get stuck sometimes, needing "hooman" help (identifying issues, sharing docs/code samples).


โ–ช ๐—–๐—ผ๐—ฑ๐—ฒ ๐—ฝ๐—ฟ๐—ถ๐˜ƒ๐—ฎ๐—ฐ๐˜† a likely challenge, but solutions are in development.


โ–ช ๐—ง๐—ต๐—ฒ๐˜† ๐—ฐ๐—ฎ๐—ป'๐˜ ๐—ฑ๐—ผ ๐—ฒ๐˜ƒ๐—ฒ๐—ฟ๐˜†๐˜๐—ต๐—ถ๐—ป๐—ด. For instance, deployments (yet. I think..)


With Devin and other full-stack bots coming up, this might all change soon. Eagerly awaiting.


๐—–๐—ข๐——๐—œ๐—ก๐—š ๐—•๐—ฌ ๐—š๐—ฃ๐—ง: ๐—˜๐—ซ๐—”๐— ๐—ฃ๐—Ÿ๐—˜๐—ฆ

Tools I've built using GPT (links in the comments). While small in scale, these tools have helped me understand GPT's application in coding, particularly in areas like LLM apps, APIs, and deployments.


๐—ฌ๐—™๐—œ๐—ก ๐—•๐—ผ๐˜๐˜€: Data transfers via FastAPI endpoints / Flowise AI Platform. Demonstrate GPT's potential for building data pipelines,ย as well as LLM app deployment within and outside GPT Store.ย 


๐—Ÿ๐—น๐—ฎ๐—บ๐—ฎ ๐—ฃ๐—ฎ๐—ฟ๐˜€๐—ฒ๐—ฟ ๐—ช๐—ถ๐—ฑ๐—ด๐—ฒ๐˜: File processing & transfers via API for LLM apps


๐—•๐˜‚๐—ถ๐—น๐—ฑ๐—ถ๐—ป๐—ด ๐— ๐—Ÿ ๐— ๐—ผ๐—ฑ๐—ฒ๐—น๐˜€ ๐˜„๐—ถ๐˜๐—ต ๐—–๐—ต๐—ฎ๐˜๐—š๐—ฃ๐—ง: Demonstrates GPT4โ€™s ability to generate ML code, as well run ML models based on prompts


๐— ๐˜‚๐˜๐˜‚๐—ฎ๐—น ๐—™๐˜‚๐—ป๐—ฑ๐˜€ ๐—”๐—น๐—น๐—ผ๐—ฐ๐—ฎ๐˜๐—ถ๐—ผ๐—ป ๐—”๐—ป๐—ฎ๐—น๐˜†๐˜‡๐—ฒ๐—ฟ (๐—š๐—ฃ๐—ง): Automation and data manipulation across multiple Excel files.


All coding by ChatGPT3.5, ChatGPT4, and Gemini ~equally. End-to-end time exponentially quicker than what I could have done alone. Free GPTs are also exceptional.


Sam Altman, "AI could boost programmers' productivity by 30X. Totally agree.

2 views
bottom of page