Deep tech,
written and visualized.

A working engineer's library: 469 in-depth articles paired with 135 single-file interactive labs on transformers, distributed systems, Apache Cassandra, agent protocols, networking, security, and real-time media. All static HTML, all yours to read offline.

★ Featured · 50+50

Transformer Math & CPU SLM

A depth-first walk through how autoregressive transformers actually work — from linear algebra to CPU inference kernels.

50 Articles
50 Labs
10 Batches
0
In-depth articles
0
Interactive labs
0
Topic categories
0%
JS framework dependency
Transformer Math Apache Cassandra MCP Protocol Raft Consensus RoPE Encoding WebRTC OAuth 2.1 SwiGLU Virtual Threads FlashAttention BBR Congestion QLoRA Kafka KRaft FlashAttention Lab HTTP/3 QUIC SmoothQuant Transformer Math Apache Cassandra MCP Protocol Raft Consensus RoPE Encoding WebRTC OAuth 2.1 SwiGLU Virtual Threads FlashAttention BBR Congestion QLoRA Kafka KRaft FlashAttention Lab HTTP/3 QUIC SmoothQuant
↑ hover to pause · click any topic to jump straight to it
Articles by topic

Hand-written technical articles,
organized by domain

Every article is self-contained, uses concrete examples and numbers, and avoids marketing fluff. Click any category to browse its index. 469 articles across 25 categories.

Advertisement
Interactive labs

No install. No build step.
135 labs ready to play.

Each lab is one HTML file with embedded JavaScript. Open it in a browser and play. Every lab has interactive controls, a clear ★ Key Takeaway, and a ▶ What to Try guide.

▶ Lab series

50 labs that visualize transformer math, end-to-end

From matrix multiplication step-through to KV cache memory growth and CPU inference latency — each lab visualizes one concept from the article series.

Read next

Start here if you're curious where to dig in

Three good starting points if you're new to the site. Each one opens onto dozens more.

About

Built by an engineer
who learns by building

Sandeep Belgavi Ashok Kumar

Belgavi's AI Lab is a personal initiative by Sandeep Belgavi Ashok Kumar, a Senior Engineering Manager and Architect working at the intersection of scalable distributed systems and modern AI.

The site captures deep technical knowledge in two complementary forms: focused written explanations and interactive single-file labs that make the underlying mechanics visible. Topics span LLM internals, agent protocols (MCP, A2A, ADK, AP2), Apache Cassandra, distributed consensus, system design, networking, security, observability, and the realtime media stack (WebRTC, HLS, DASH).

Every article aims for concrete numbers and working examples over marketing claims. Every lab runs in a browser — no install, no build step, no JavaScript framework dependency.