asopi tech
asopi techIndie Developer
Blog

Field notes & research

41 articles ・ 25 tags

Research notes, implementation tips, and operational insights. Searchable by tags and keywords.

View

Categories

Tags

41 articles

25 tags

Consensus Algorithms, Underground (Part 4) — A World Where Identities Cannot Be Counted: PoW and PoS
DATA
Data & DB

Consensus Algorithms, Underground (Part 4) — A World Where Identities Cannot Be Counted: PoW and PoS

In a network where neither the identity nor the number of participants is known, one node one vote does not hold. Probabilistic finality in PoW, the structure by which PoS relies on economic penalties rather than impossibility, Solana and Avalanche, and the option of avoiding agreement — with the criteria for choosing among them. Part 4 of a 4-part series.

#data#distributed
Consensus Algorithms, Underground (Part 1) — Crashes and Delays Are Indistinguishable
DATA
Data & DB

Consensus Algorithms, Underground (Part 1) — Crashes and Delays Are Indistinguishable

A node that stops responding may have crashed, or it may merely be slow; the two cannot be told apart. This part separates leader election, ordering, and commitment into three distinct problems, and covers the practical meaning of FLP impossibility, split brain and fencing tokens, and the boundary between the replicated log and the applied state. Part 1 of a 4-part series.

#data#distributed
Even So, I'm Building Alopex DB
ALOPEX
Alopex DB

Even So, I'm Building Alopex DB

SQLite, DuckDB, PostgreSQL, distributed DBs, vector DBs, graph DBs — there is no shortage of great products. I'm building a new database anyway because every time the project moves forward, I re-pick the DB and rebuild the data itself. Alopex DB aims to be a foundation that keeps data's volume, conversion time, regeneration cost, and provenance intact as you scale from local to cloud, from single node to distributed.

#data#ai#sql#wsl
Part 5: Toward Physically Correct Video — World-Model Integration and the Era of AI-Made Training Data
AI
GenAI

Part 5: Toward Physically Correct Video — World-Model Integration and the Era of AI-Made Training Data

Plain video generation tends to produce footage that "looks natural but is physically wrong": objects vanish, gravity is ignored, and the results of actions aren't consistent. Part 5 (the finale) shows, with minimal code, how to embed a world model into video generation and put physical constraints into the loss to make "physically consistent video" — and paints the future of building training data wholesale with AI and simulation. Part 5 of a 5-part series.

#ai#physical-ai#nim
Part 4: What's the Difference Between Video Generation and World Models? — "Plausible Footage" vs. "State Transitions"
AI
GenAI

Part 4: What's the Difference Between Video Generation and World Models? — "Plausible Footage" vs. "State Transitions"

Video generation like Sora and Veo looks similar to world models but is a different thing. The former makes "plausible footage"; the latter learns "how the world changes when you act." Part 4 contrasts video generation's noise prediction with a world model's `S_t + A_t → S_{t+1}`, and clarifies what NVIDIA Cosmos and Google Genie use as training data. Part 4 of a 5-part series.

#ai#physical-ai#robotics
Part 3: Speech Models Converge Toward LLMs — TTS and Speech2Speech Training Data
AI
GenAI

Part 3: Speech Models Converge Toward LLMs — TTS and Speech2Speech Training Data

Text-to-speech once used the Mel spectrogram as its answer. But recent models convert audio into "audio tokens" and predict those — a structure that now looks just like an LLM. Part 3 follows how training data is built for TTS and Speech2Speech (speech translation, voice conversion, end-to-end conversational AI), including the difficulty of gathering paired data and how it's worked around. Part 3 of a 5-part series.

#ai#learning
Part 2: Image Generation Guesses the "Noise" or the "Token" — TXT2IMG Training Data and Captions
AI
GenAI

Part 2: Image Generation Guesses the "Noise" or the "Token" — TXT2IMG Training Data and Captions

Image generation comes in lineages — GAN, autoregressive (token) type, and diffusion — and today's mainstream is diffusion models like Stable Diffusion. The diffusion type learns to "guess the added noise" rather than "draw the image directly," while the autoregressive type guesses the "next image token" like an LLM. Part 2 covers gathering image-text pairs, CLIP quality filtering, and the training data for both approaches — with minimal code. Part 2 of a 5-part series.

#ai#stablediffusion#learning
Part 1: What Does Generative AI Learn as the "Answer"? — Self-Supervised Learning and LLMs
AI
GenAI

Part 1: What Does Generative AI Learn as the "Answer"? — Self-Supervised Learning and LLMs

Neither LLMs nor image generators learn from training data that humans have labeled item by item. The key is self-supervised learning: mechanically constructing inputs and answers from raw data. Part 1 clarifies the difference between raw data and training data, and shows how an LLM turns text itself into a next-token prediction problem, with minimal code. Part 1 of a 5-part series.

#ai#learning#data
Part 3: Don't Do Everything with One Giant Model — An Architecture of Hierarchy and Separated Responsibilities
AI
GenAI

Part 3: Don't Do Everything with One Giant Model — An Architecture of Hierarchy and Separated Responsibilities

Billions of parameters fire just to decide "I'm hungry"—doing everything with a single giant model is a waste of resources. Just as the human nervous system splits circuits by type of processing, AI should be layered by time scale and responsibility. Robot control hierarchies, energy efficiency, an OS-like structure, and why the control software itself must be organized. The finale of a 3-part series.

#ai#robotics#physical-ai
Part 2: Give AI a "Cerebellum" and "Autonomic Nerves" — A Blueprint for Embodied Intelligence
AI
GenAI

Part 2: Give AI a "Cerebellum" and "Autonomic Nerves" — A Blueprint for Embodied Intelligence

If we liken the text-grown LLM to the human cerebrum, there should also be equivalents of the cerebellum and spinal cord, and of the autonomic nerves that govern physiological feedback from the heart and gut. From a body-to-AI layer mapping to the artificial cerebellum (Diffusion Policy), interoception, and homeostasis, this post draws a blueprint for embodied intelligence. Part 2 of a 3-part series.

#ai#robotics#physical-ai
Part 1: Where Physical AI Stands Now — An Extension of LLMs, or Something Else?
AI
GenAI

Part 1: Where Physical AI Stands Now — An Extension of LLMs, or Something Else?

Physical AI unifies seeing, understanding language, and physically acting in a robot. Is it an outgrowth of LLM research, or a separate lineage like diffusion models? From the three lineages—VLA, diffusion/flow, and world models—and concrete models like RT-2, Gemini Robotics, OpenVLA, π0, GR00T, and Cosmos, we map where things stand as of 2026. Part 1 of a 3-part series.

#ai#robotics#physical-ai
US vs. Japan IT Industries — "Product" and "Service," and the AI Reshuffle
AI
GenAI

US vs. Japan IT Industries — "Product" and "Service," and the AI Reshuffle

In the US, an "IT company" means Microsoft or NVIDIA; in Japan, it means NTT Data or Fujitsu. The same words point to different things. One side mass-produces products for the world; the other supports each customer's bespoke operations. That structural gap has split revenue, talent, and competitiveness. How does generative AI reshape it? A look from Japan's weaknesses and strengths.

#ai#industry#career
The Skill AI Can't Beat: Data Engineering
DATA
Data & DB

The Skill AI Can't Beat: Data Engineering

Coding agents are strong at logic and tests. But they can't tell what a piece of data means, who owns it, how fresh it flows, or which copy is authoritative. It follows from the fact that today's AI has no embodiment: the people who rise in value are those who can design the meaning, quality, lineage, and responsibility of data. A look at the trend with the latest data-engineering discussion.

#data#ai#career
Rust Is Not a Silver Bullet — Language Parsers Belong in Nim or Roc
DEV
Engineering

Rust Is Not a Silver Bullet — Language Parsers Belong in Nim or Roc

The myth that 'Rust can do anything' collapses the moment you try to scale a recursive AST. For SQL parsers — massive tagged unions, deep recursion — Rust's type system breaks down in both compile time and maintainability. A 3-language benchmark porting the same parser to Nim and Roc shows exactly why Rust is structurally the wrong tool, with runnable code.

#nim#roc#rust#sql

The Responsibilities MCP Separates, Part 2 — Separating Identity from Permission to Data

MCP v2026-07-28 uses Bearer access tokens to establish the requester at the router and to decide whether that requester may operate on business data.

The Responsibilities MCP Separates, Part 1 — Moving Protocol State Out of the Session

MCP v2026-07-28 separates protocol state from application state that once shared a connection, allowing any instance to process a request. A visual guide to the fields and flow.

Where Agent-Initiated Payments Are Heading — The Layers Behind AP2, ACP, UCP and x402

It was never about handing a card number to a model. So what are all these new standards actually trying to prove?

The Skill of Not Building in the AI Agent Era (5) — Six Checks to Run Before You Start

Will that code still be needed once the next model ships? Six things worth checking before you start.

The Skill of Not Building in the AI Agent Era (4) — Implementations That Don't Repay the Effort, and the Case for Waiting

Homegrown long-term memory, a workflow DSL, a model router. Each one works, and each is genuinely interesting to build. But what if three months of waiting made them unnecessary?

The Skill of Not Building in the AI Agent Era (3) — Run the Loop Instead of Writing by Hand

Boris Cherny, who built Claude Code, says he hasn't hand-written a line of code in eight months. Is the craft of polishing prompts giving way to the craft of designing loops?

The Skill of Not Building in the AI Agent Era (2) — Implementations at the Platform's Mercy

The code runs and the tests pass. Only the value of using it has vanished. A look at this failure mode as a stranded asset rather than technical debt.

The Skill of Not Building in the AI Agent Era (1) — Third Parties Swallowed by the Platform

PDF chat, AI slides, ChatGPT plugins: all standard features of ChatGPT and Gemini now. Where did the implementations that once powered them go?

MoonBit and the State of WASM Languages (2026) — Choosing a Language in the Wasm GC Era

Comparing WASM-first languages (MoonBit, AssemblyScript, Grain) with languages that added a WASM backend (Rust, C/C++, TinyGo, Kotlin/Wasm, Dart, C#, SwiftWasm). Using linear memory vs. Wasm GC, Component Model support, binary size, and maturity as axes, this article explains how to choose a WASM language in 2026.

A Map of WASM Runtimes and Frameworks (2026) — Who Actually Implements the Raw APIs?

From WASM runtimes like Wasmtime, Wasmer, WasmEdge, wazero, and WAMR, through WASI/Component Model tooling (WIT, wit-bindgen, cargo-component, jco), upper frameworks (Extism, Spin, wasmCloud, Proxy-Wasm), to GPU implementations (Dawn, wgpu) — a layer-by-layer map of who implements WASM's raw APIs, and how to choose, as of 2026.

WASM, WebGL, and WebGPU in 2026 — Not Replacement, but Division of Roles

A 2026 survey of WebAssembly, WebGL, and WebGPU — plus WASI/Component Model, Web Workers, WebCodecs, and WebNN. Rather than a simple 'JS→WASM' or 'WebGL→WebGPU' replacement, the web platform has converged on a per-layer division of roles. This article maps the current state and offers guidance on technology choice and architecture.

Deep Insights into Nim Syntax — How Similar Is It to Pascal, Python, and Haskell?

A 38-topic analysis of Nim syntax — declarations, the type system, expression orientation, generics, distinct types, macros, and compile-time execution — comparing how closely each construct resembles the styles of Pascal, Python, C-family, and functional languages, with a primer on block syntax design and the off-side rule.

What Is the Best Database for a Startup? — A Stage-by-Stage Use-Case Analysis and the 'Scaling Cliff'

An analysis of how database requirements change across the MVP, PMF, growth, and scale stages, comparing the trade-offs of SQLite, Postgres, managed databases, and dedicated vector stores — and where Alopex DB's 'erase the migration cliff' approach fits.

What Alopex Chirps Is — A Library for Forming Clusters, and the Consensus It Adopted

Building a distributed system means building node discovery, liveness detection, messaging, and agreement before getting to the actual work. Alopex Chirps takes on exactly those four. This covers what it handles and what it does not, why QUIC was chosen, and how openraft came to be the consensus layer.

Consensus Algorithms, Underground (Part 4) — A World Where Identities Cannot Be Counted: PoW and PoS

In a network where neither the identity nor the number of participants is known, one node one vote does not hold. Probabilistic finality in PoW, the structure by which PoS relies on economic penalties rather than impossibility, Solana and Avalanche, and the option of avoiding agreement — with the criteria for choosing among them. Part 4 of a 4-part series.

Consensus Algorithms, Underground (Part 3) — Lying Nodes, from PBFT to DAG-BFT

The world in which nodes send signed, contradictory information. The meaning of the constant n=3f+1, PBFT's three-phase voting, Tendermint's locking rules, HotStuff's quorum certificates, and what the DAG-BFT designs since Narwhal separated and what they removed. Part 3 of a 4-part series.

Consensus Algorithms, Underground (Part 2) — The Crash Failure Family: Paxos and Raft

The family of algorithms for the crash failure model, where nodes stop but do not lie. From the single tool Paxos relies on — quorum intersection — through Raft's decomposition, EPaxos, Multi-Raft, and the parts outside the papers: persistence ordering and linearizable reads, with minimal code. Part 2 of a 4-part series.

Consensus Algorithms, Underground (Part 1) — Crashes and Delays Are Indistinguishable

A node that stops responding may have crashed, or it may merely be slow; the two cannot be told apart. This part separates leader election, ordering, and commitment into three distinct problems, and covers the practical meaning of FLP impossibility, split brain and fencing tokens, and the boundary between the replicated log and the applied state. Part 1 of a 4-part series.

Getting Started with Nim: Tauri-like Nimino and Zed-like Nimculus

A quick introduction to Nim, and an announcement of two desktop products I started in Nim — Nimino, a Tauri-like lightweight WebView app framework, and Nimculus, a Zed-like GPU-native code editor.

Finding a Local Japanese LLM That's Good at Writing Blog Posts

I picked one local Japanese LLM to write this blog, from the ones that run on my M3 MacBook Air (24GB). Here are the results of comparing them on long-form Japanese quality, plus the exact Ollama API calls and the prompts I used.

Even So, I'm Building Alopex DB

SQLite, DuckDB, PostgreSQL, distributed DBs, vector DBs, graph DBs — there is no shortage of great products. I'm building a new database anyway because every time the project moves forward, I re-pick the DB and rebuild the data itself. Alopex DB aims to be a foundation that keeps data's volume, conversion time, regeneration cost, and provenance intact as you scale from local to cloud, from single node to distributed.

Part 5: Toward Physically Correct Video — World-Model Integration and the Era of AI-Made Training Data

Plain video generation tends to produce footage that "looks natural but is physically wrong": objects vanish, gravity is ignored, and the results of actions aren't consistent. Part 5 (the finale) shows, with minimal code, how to embed a world model into video generation and put physical constraints into the loss to make "physically consistent video" — and paints the future of building training data wholesale with AI and simulation. Part 5 of a 5-part series.

Part 4: What's the Difference Between Video Generation and World Models? — "Plausible Footage" vs. "State Transitions"

Video generation like Sora and Veo looks similar to world models but is a different thing. The former makes "plausible footage"; the latter learns "how the world changes when you act." Part 4 contrasts video generation's noise prediction with a world model's `S_t + A_t → S_{t+1}`, and clarifies what NVIDIA Cosmos and Google Genie use as training data. Part 4 of a 5-part series.

Part 3: Speech Models Converge Toward LLMs — TTS and Speech2Speech Training Data

Text-to-speech once used the Mel spectrogram as its answer. But recent models convert audio into "audio tokens" and predict those — a structure that now looks just like an LLM. Part 3 follows how training data is built for TTS and Speech2Speech (speech translation, voice conversion, end-to-end conversational AI), including the difficulty of gathering paired data and how it's worked around. Part 3 of a 5-part series.

Part 2: Image Generation Guesses the "Noise" or the "Token" — TXT2IMG Training Data and Captions

Image generation comes in lineages — GAN, autoregressive (token) type, and diffusion — and today's mainstream is diffusion models like Stable Diffusion. The diffusion type learns to "guess the added noise" rather than "draw the image directly," while the autoregressive type guesses the "next image token" like an LLM. Part 2 covers gathering image-text pairs, CLIP quality filtering, and the training data for both approaches — with minimal code. Part 2 of a 5-part series.

Part 1: What Does Generative AI Learn as the "Answer"? — Self-Supervised Learning and LLMs

Neither LLMs nor image generators learn from training data that humans have labeled item by item. The key is self-supervised learning: mechanically constructing inputs and answers from raw data. Part 1 clarifies the difference between raw data and training data, and shows how an LLM turns text itself into a next-token prediction problem, with minimal code. Part 1 of a 5-part series.

Part 3: Don't Do Everything with One Giant Model — An Architecture of Hierarchy and Separated Responsibilities

Billions of parameters fire just to decide "I'm hungry"—doing everything with a single giant model is a waste of resources. Just as the human nervous system splits circuits by type of processing, AI should be layered by time scale and responsibility. Robot control hierarchies, energy efficiency, an OS-like structure, and why the control software itself must be organized. The finale of a 3-part series.

Part 2: Give AI a "Cerebellum" and "Autonomic Nerves" — A Blueprint for Embodied Intelligence

If we liken the text-grown LLM to the human cerebrum, there should also be equivalents of the cerebellum and spinal cord, and of the autonomic nerves that govern physiological feedback from the heart and gut. From a body-to-AI layer mapping to the artificial cerebellum (Diffusion Policy), interoception, and homeostasis, this post draws a blueprint for embodied intelligence. Part 2 of a 3-part series.

Part 1: Where Physical AI Stands Now — An Extension of LLMs, or Something Else?

Physical AI unifies seeing, understanding language, and physically acting in a robot. Is it an outgrowth of LLM research, or a separate lineage like diffusion models? From the three lineages—VLA, diffusion/flow, and world models—and concrete models like RT-2, Gemini Robotics, OpenVLA, π0, GR00T, and Cosmos, we map where things stand as of 2026. Part 1 of a 3-part series.

US vs. Japan IT Industries — "Product" and "Service," and the AI Reshuffle

In the US, an "IT company" means Microsoft or NVIDIA; in Japan, it means NTT Data or Fujitsu. The same words point to different things. One side mass-produces products for the world; the other supports each customer's bespoke operations. That structural gap has split revenue, talent, and competitiveness. How does generative AI reshape it? A look from Japan's weaknesses and strengths.

The Skill AI Can't Beat: Data Engineering

Coding agents are strong at logic and tests. But they can't tell what a piece of data means, who owns it, how fresh it flows, or which copy is authoritative. It follows from the fact that today's AI has no embodiment: the people who rise in value are those who can design the meaning, quality, lineage, and responsibility of data. A look at the trend with the latest data-engineering discussion.

Part 5: Those Who Grow AI Win — From "Users" to "Growers"

The finale. The core skill common to every part is turning experience into a form both people and AI can act on. Why it does not commoditize and instead compounds, the daily practices, and even "what if AI gains a body?"—the synthesis goes deep.

Part 4: Grow Leaders and Managers with an AI Team — Directing Is Developing

Break work down, share expectations and criteria, give feedback. The template of good management overlaps exactly with good instructions to an AI team. How the experience of leading AI grows leaders and managers, and how it turns the organization's learning loop.

Part 3: Make AI Your Student and Escape the Intermediate Trap — Graduating from "I Thought I Understood"

The key for intermediates who stall at "I basically get it" is to teach AI. Verbalizing your premises, constraints, criteria, and edge cases becomes training that turns tacit knowledge into explicit knowledge. A learning method that uses AI as a mirror of your understanding, plus a practical five-step routine.

Part 2: Beginners, Build Experience with AI — The Ones Who Grow Fastest Are Beginners

In an era where AI hands you answers, are beginners at a disadvantage? Quite the opposite. By trying sticky material—dev environments, containers, VMs—together with AI and failing together, beginners build experience the fastest through iteration. Includes how to let AI fail a lot and learn by watching. Part 2.

Part 1: How Not to Let AI Take Your Career — A Career Path for the AI Era

Many people worry that AI will take their jobs. But reframing the question opens a path forward. As the introduction to a five-part series, this piece maps out AI-era careers across the beginner, intermediate, and leader stages.

Rust Is Not a Silver Bullet — Language Parsers Belong in Nim or Roc

The myth that 'Rust can do anything' collapses the moment you try to scale a recursive AST. For SQL parsers — massive tagged unions, deep recursion — Rust's type system breaks down in both compile time and maintainability. A 3-language benchmark porting the same parser to Nim and Roc shows exactly why Rust is structurally the wrong tool, with runnable code.

Getting Started with Roc: A Pure Functional Language Meant to Ride Alongside Rust

A technical look at Roc — Richard Feldman's pure functional language. Covering the alpha4-rolling status, Perceus reference counting, the Platform/Host model, and the Zig compiler rewrite, informed by AlopexDB's SQL parser trial.

Getting Started with Nim: A Multi-Paradigm Language That Transpiles to C

A technical overview of Nim in 2026 — the stable 2.2 series, the upcoming 3.0 (Nimony) compiler rewrite, ORC memory management, C ABI interop, and its real-world performance as an SQL parser candidate for AlopexDB.

Astro Features Used in the Site Redesign: A Technical Deep Dive

A technical explanation of the Astro features used in the Asopi Tech site redesign. Covering Content Collections, i18n support, dynamic routing, OG image generation, and more.

Major Site Redesign: Introducing the New Site Structure and Pages

We've completely redesigned the Asopi Tech website. Introducing the new site structure with multilingual support, OSS project pages, and service landing pages.

No posts found

Try changing your filters or search keywords.