Daniel Obiorah (@DannyThePyDev) header
Daniel Obiorah (@DannyThePyDev) profile picture

Backend and AI Engineer. Github: https://teapawt.net/t.co/2yISEUnC9n Building https://teapawt.net/t.co/SegBl401fu

Kaduna, Nigeria
Joined January 2023
Photo by Daniel Obiorah (@DannyThePyDev)Photo by Daniel Obiorah (@DannyThePyDev)Photo by Daniel Obiorah (@DannyThePyDev)Photo by Daniel Obiorah (@DannyThePyDev)Photo by Daniel Obiorah (@DannyThePyDev)Photo by Daniel Obiorah (@DannyThePyDev)Photo by Daniel Obiorah (@DannyThePyDev)Photo by Daniel Obiorah (@DannyThePyDev)Photo by Daniel Obiorah (@DannyThePyDev)Photo by Daniel Obiorah (@DannyThePyDev)
Sadly as a Nigerian, dem no go rate you. Very frustrating stuff tbh. The money is there but they'll hoard it. They prefer to fund gimmicks than what's real, hype train riders. At best you'll get "skill empowerment and mentorship" 😂
how are people raising funding rounds?? how did you guys meet them?? bro, i’m genuinely tired, i’ve tried building in public, linkedin, literally everywhere omo.
20
Full-stack dev kit for N-ATLAS • Python SDK & CLI: natlas setup + natlas serve for local inference • TypeScript SDK (@natlas/sdk): Zero-dep • Anti-Drift Tone Guards: • Direct Audio Pipelines: system ffmpeg Repo: github.com/Beetroit/NAtlas Would love your feedback and stars! ⭐️
The N-ATLAS Innovation Challenge is calling on Nigerian developers, researchers, entrepreneurs and innovators to transform the platform into practical, accessible solutions that people can use in their everyday lives. From developer tools and voice-first applications to sector-specific AI solutions addressing real needs in healthcare, agriculture, education, financial services and other critical sectors. Whether you are from academia, the private sector or government, this is an opportunity to collaborate and demonstrate how Nigerian talent can build locally relevant AI solutions at scale on Nigeria’s Large Language Model (N-ATLAS) You can find more details here - ncair.nitda.gov.ng/naic/ Application deadline is October 12th, 2026. Build on N-ATLAS, for Nigeria. @NCAIRNigeria @ONDINigeria @FMCIDENigeria
4
1
1
7
188
PRs are welcome, raise issues and hell, i'm down for all that.🤭
29
Proof of effort.
Really really hate losing.
1
128
Builders🌝💪🏾
4 Days ago we were at the Superteam Co-Hack event with @SuperteamNG Kaduna builders, we've embarked on an inspiring journey of collaboration and innovation. From idea generation to solution development, we've pushed the boundaries of Web3, solving challenges...
1
12
196
Daniel Obiorah retweeted
Need a private office or auditorium for your next event? Look no further! Almara has you covered. Visit us today at No. 22b kanta Road Kaduna City Center or DM us for more information. #Almara #EventSpace #Kaduna #OfficeSpace #Auditorium
5
8
482
Building @superteam CoHack Event with @DannyThePyDev @kp_creativ 🚀🫡🚀
I'm live at the @superteam CoHack event with @SuperteamNG Kaduna builders 🔥💪
3
11
286
you can't out-train bad data!
3
2
1
12
1,226
Wait till you discover ALT + SHIFT + UP/DOWN ARROWS CTRL + SHIFT + UP/DOWN ARROWS 🙃
turns out multi-cursor editing also works while using jupyter notebooks in vscode. to do this, press and hold alt and then click on the lines where you want to add cursors. (for jupyter lab or notebooks, press and hold ctrl)
2
125
If the truth will hurt the person at the time, defer. Don't lie, but say a different truth. You can speak your mind later.
There's a difference between authenticity & word vomit I will say whatever I want whenever I want & I don't say everything I ask myself "Will it have a positive or a negative impact on me? or someone I care about?" Be wary of those that use authentic to be unkind or uncaring
1
96
And oh, gemini 1.5 pro is my second best LLM. Claude still tops, especially when you want strictly json responses without the unnecessary "certainly,..". Used it to automate text clustering for columns with non sensitive data.
1
142
Takes around 20 secs to convert 1. Will just leave the PC running overnight. Had to parallelize it sha, so I can delve into other tasks.
58
Spent a whole day searching for a way to convert txt to pdf in python, not one-liners, complete 300+ pages of text, these LLMs were just recycling nonsense. Found it in the docs after like 5 minutes of research.😒
1
53
Was contracted to clean up one of the dirtiest survey datasets I've seen so far. They wanted to convert many freeform text columns into categories, 10 max per column. With over 200 unique values, I had to use a dict and the Series.str.contains method. Real helpful.
43
With each completed gig, i learn something new. That's on very fun part of freelancing, each client's requirement is different from the next, even though there may exist similarities.
40
Before cleaning vs After num_duplicates : num_samples : num_unique I can sleep now.🤧
1
47
I once crashed my vscode by printing 500k lines of a csv.😅 Now i use break excessively when looping. That way, i don't have to rewrite the logic when i'm done testing. I typically use indexing for selenium elements. Peep my hack for indexing a list of dicts with a dict btw. 😅
3
115
You think you are good at data cleaning and preprocessing until you come face to face with a very screwed up annotated image data. My colab notebook was so messy i wouldn't touch it for days. Currently done cleaning, 3am motivation ftw i guess.😅
1
1
2
59
Found it very interesting that there's a large number of data nerds who do no know vscode supports ipynb. One kept insisting that jupyter was better, bruhh.😅
Replying to @maureenondata
I will recommend vs code Reasons: U can run both ipynb and py file on it Also U have access to many extensions that will make Ur work easier and faster U can easily arrange your files, delete them and also use version control. But you need to set it up first and understand it
1
2
114