Hey, I’m arves

I’m a software engineer who likes building tools and systems that feel clean, fast and useful.

Over the past year, I’ve been deep into text search and retrieval.

I work mostly in Python for its elegance, clarity and immense versatility.

Projects


Retrievvy [GitHub]

Fast search for texts

Most people go straight to massive embeddings models or expensive fine-tuning. I didn’t.

Retrievvy uses a tiny, fast embeddings model alongside classic keyword search. Lightweight by design.

Dense search handles the semantics. BM25 catches the specific terms you don’t want to miss.

Results from both are combined using a reranking algorithm I wrote to bring out the best of both worlds.

The result? Fast, meaningful search that’s simple to deploy and actually works.

Notes




Design systems to do one thing exceptionally well, then make them easy to extend externally.

Don’t abstract until it hurts, and you understand the pain.

Favor simple designs, but not at the cost of power or flexibility.