en
Feedback
SnackShell

SnackShell

Open in Telegram

Some AI and ML Stuffs. Journey to AI & ML Engineer | snackshell.work Founder @BanaCodes & @SelamGPT

Show more
The country is not specifiedThe category is not specified
1 246
Subscribers
-224 hours
-157 days
-9430 days
Posts Archive
my fav dopamine :)

Repost from N/a
SelamGPT works smoothly across multiple languages with automatic, natural conversation flow. It even understands transliterat
+3
SelamGPT works smoothly across multiple languages with automatic, natural conversation flow. It even understands transliteration. like Amharic or Tigrinya written in Latin letters and so much more.

Repost from N/a
Native Ethiopian AI. Built for Builders The wait is over. platform.selamgpt.com is officially live. ​We’ve moved past the inf
+6
Native Ethiopian AI. Built for Builders The wait is over. platform.selamgpt.com is officially live. ​We’ve moved past the infrastructure challenges to deliver a production-ready environment designed for developers. ​What’s inside: βœ… OpenAI Compatible: seamless integration with your existing codebase. βœ… Native Support: LLMs & TTS optimized for Amharic, Tigrinya, and Oromo. βœ… Reliable: Fully operational status with transparent monitoring. ​​A note to our builders: As this is the earliest stage of our public platform, you may encounter occasional bugs or rough edges. We are constantly refining the system, and your input helps us get there faster. πŸ“§ Send any issues or suggestions to: feedback@selamgpt.com ​The platform is yours now. ❀️ ​Start building: https://platform.selamgpt.com

So sad I missed Dagi's meetup. I couldn't make it there. Maybe next time, God willing.

Repost from Bana Codes
So sad I missed the meetup. I couldn't make it there. Maybe next time, God willing.

I need some coffee πŸ”«
+1
I need some coffee πŸ”«

ObsidianUI β€” Built to ship. 50+ professional, enterprise-ready components Try it β†’ https://obsidianui.dev Open-Source Fast. Local. Free forever. Source code included. No bullshit. https://github.com/Atharvsinh-codez/ObsidianUI

I was fixing some bugs today. I hope I fixed all the bugs that was forgotten and unseen in the entire selamgpt backend codes. Tomorrow I'll be adding file upload (pdf, text, docs..) and OCR features. I hope I'll finish and deploy the backend to my VPS and finish the playground and docs with all the new features I have been adding. After this I'll be moving to Making the chat app. This will be the final one for launching. Hope I'll be done soon.

When the code finally lets you breathe. Just taking a break
When the code finally lets you breathe. Just taking a break

noted.

The kinds of messages I've been getting into the SelamGPT Waitlist contact form mail. πŸ˜‚ There are people who don't know what
The kinds of messages I've been getting into the SelamGPT Waitlist contact form mail. πŸ˜‚ There are people who don't know what a Waitlist is. Anyways, in this week I've been working adding a Websocket connection feature for our selam tts models we will have across the 4 Ethiopian languages + English tts, including gpt-audio models into our selam api platform. Almost finishing the tts and voice agent api. Next phase will be integrating the backend with the frontend platform and finalising the databases and auth stuffs.

and yeah. Marry Christmas to ALL Ethiopian Fams. πŸ‡ͺπŸ‡Ή

Everything is integrated and ready for deployment. Also the tts is done and hosted, but I have to add the tts api into SELAMAI backend. Hopefully we're going to launch very soon.

Current backend status. Normal chat βœ”οΈ Chat with streaming βœ”οΈ Tool/function calling βœ”οΈ Batch processing βœ”οΈ Web search βœ”οΈ Image generation βœ”οΈ Image editing βœ”οΈ Live Voice agent βœ”οΈ TTS services βœ”οΈ Supported Languages With this all features: Amharic, Tigrinya, Oromo, Somali, Afar, English.

Done working on the tts and voice agent. Now integrating it with selam api. Current supported tts and voice agent languages. Amharicβœ”οΈ Tigrinya βœ”οΈ Oromo βœ”οΈ English βœ”οΈ Somali βœ”οΈ Afar ❌ (not yet)

Btw, this is not the actual LLM, it's gonna be much native and fulent. This is just Google's model with system prompt.

I let my little sister talk to the voice agent I made for testing to add into our selamg api, and this is what it responds. "αˆ…α‹­α‹ˆα‰΅αˆ² ከም α‰£αˆ•αˆͺ αŠ₯α‹«" πŸ˜‚