sankode and sankript : a devanagrai native programming language

If you have read my previous posts from years ago (c.f. https://blog.tovganesh.in/2007/08/programming-in-and-for-devanagari.html) I have always had an interest in building a language that is devanagri first and explore the possibilities of how it will be. So I eventually wrote a spec and with help of Google Antigravity built the first working version of the a new programming language (sankode) and a scripting language called (sankript) with a very basic IDE to play around, and a basic tutorial with examples. 
I also built a basic LLM, and a ReLU based LLM using entirely sankode and used abhigyan shakuntalam as a training set. It has been so much fun making this, try it out and let the feedback flow in. 
In the mean time I also encountered Laya developed by Nandakishor M (Nandakishor m (@Nandakishorm1) / X) - so I tried out build that as well using sankode - it seems to work ok.

                                 sankode studio: a simple IDE to play around with sankode and sankript                            

                                  sankode laya

Back after a lull in writing

The past 3 year have been rather enjoying for me in my learning and life. Technically have learned and built enterprise grade air-gapped RAG systems, helped co-found a company HouzeGuru - which is building digital tools in the form of a super app for Indian Families and learned to custom train an LLM model from scratch. 

But the most fun things have been to emply coding agents to do things that I always wanted to do but did never get the bandwidth to do - build an OS, and build a language that is Devanagari native. Both of these are my currently going pet projects : check kosh (https://github.com/tovganesh/kosh)  and sankode ( https://github.com/tovganesh/sankode ) repos.

In the meantime - I didn't add much in my personal hardware. The 32gb windows machine has been the best buy for all my works. I had access to a Mac mini M4 base model from an year now - and I got it just in time before the pricing went crazy northwards. I just do not think any hardware is worth buying for at-least next two years. Their expense to usefulness has significantly deteriorated - unless you just have too much money to put in. Oh and if you have too much money - you may as well think of sponsoring my open source projects - which are for public good - and research. 

More on what I have been building in the past months and more in up coming posts.