HN user

ashvardanian

4,504 karma

Bootstrapping <Unum.cloud> since 2015. Investing at <AAL.vc>. Hosting ML/Systems hackathons around the globe. Living between London, San Francisco, and Yerevan. Converting outrageous amounts of coffee into modest amounts of Assembly for Intel, Arm, and Nvidia chips. Maintaining FOSS libs deployed on ≫ 100M devices.

<ashvardanian.com>

<github.com/ashvardanian>

Posts387

Comments0

Using AI to build your own software

Show HN: ForkUnion v3 – Faster OpenMP-style thread-pool for C, C++, Rust, & Zig

AI Model Co-Design: Hardware-Friendly LLM Design

KinetIQ Ascend: Toward 100% Reliable Manipulation and Superhuman Speed

Nvidia CUDA Python 1.0 and CUDA 13.3 Release

Laguna XS.2 and Laguna M.1 by Poolside

FP8 Search and KV-Caching in USearch

Escaping the Fork: How Meta Modernized WebRTC Across 50 Use Cases

Porting Go's io package to C

Schema as the Core of Reliability in AI Memory

NumKong: 2'000 Mixed Precision Kernels for All

The State of Allocators in 2026

Recraft V4

Nebius to buy AI agent search company Tavily for 275M

Running Async WebAssembly on Seastar's Reactor

Open source USearch library jumpstarts ScyllaDB vector search

Mixedbread: How We Built Multimodal Late-Interaction at Billion Scale

Nvidia: Using Context as Training Data Unlocks Models That Learn at Test-Time

Nvidia Kicks Off the Next Generation of AI with Rubin – Six New Chips

Building Code-Chunk: AST Aware Code Chunking

Rust: Using String and andstr in your APIs

From Ts_rank to BM25. Introducing Pg_textsearch: True BM25 Ranking and Retrieval

Full Unicode Search at 50× ICU Speed with AVX‑512

Emulating AVX-512 intrinsics in Miri

Why we built Lightpanda in Zig

Nvidia cuTile: Python DSL and a new IR for tile-based CUDA kernels

SereneDB Secures $2.1M to Challenge the Status Quo of Search and Analytics

Evidently 0.7.17: open-source LLM tracing and dataset management

How Good Are Chinese CPUs? Benchmarking the Loongson 3A6000

Symmetric Power Transformers