01
Ship measurable impact
Performance gains, reliability, business outcomes. I aim for changes that move a number — like the 20–25% backend gains I shipped on recent SaaS work.
✓Kaito Ota
open to workSenior Fullstack & AI Engineer · 8+ years shipping production AI
Nagoya, Japan · JST (UTC+9) · usually online 09–19 JST
Senior Fullstack & AI Engineer with over 8 years of experience designing, building and deploying advanced AI systems. Graduated from Niigata University, now leading a small cross-functional team in Japan that covers AI engineering, fullstack development and go-to-market execution. Cross-industry work across healthcare, finance, education and SaaS — merging AI, software engineering and business insight to deliver measurable impact.
8+ yrs
experience
15+
projects
4
companies
2
languages
/ now
last updated · today- ◈BuildingAI-driven SaaS & automation as a freelancer
- ✦Working withOpenAI · LangChain · PyTorch · vector DBs
- ✉AvailableContract & advisory engagements
- ⌁Based inNagoya, Japan · open to remote
How I work
Three principles, learned across SaaS, automation, and enterprise work.
02
AI where it earns its place
Chatbots, NLP, intelligent search, workflow automation — used where they replace real friction, not as decoration on top of a working product.
03
Full stack, end to end
Comfortable from contracts and infra up to React UI. Most projects need someone who can hold the whole picture; I like being that person.
Languages
- Japanese· Native
- English· Proficient
8+ years · 4 companies
Where I've been
Reverse-chronological. Drawn directly from my CV.
- FR09/2022 — Present
Senior AI Engineer · Freelance
Nagoya, Japan · Remote
Delivered 10+ full-stack web and AI-powered applications for startups and business clients across SaaS, automation and Web3-related projects.
- Built scalable backend APIs, database systems and responsive frontend applications using modern JavaScript frameworks and Python-based services.
- Integrated AI-driven features such as chatbot systems, text processing, workflow automation and intelligent search into production applications.
- Improved API and application performance by approximately 20–25% through backend optimization and database query improvements.
- Built secure business applications with authentication, payment integration and third-party API connectivity.
- Managed deployment, monitoring and maintenance of cloud-hosted applications while collaborating with remote teams and clients.
TypeScriptNext.jsPythonFastAPIOpenAI APILangChainPostgreSQLAWS - CB10/2020 — 06/2022
AI & Full Stack Developer · CBTW IT & Technology / Positive Thinking Company
Remote
Developed and maintained 5+ enterprise web applications with scalable backend services and modern frontend interfaces.
- Implemented AI-related functionality including automation tools, text-processing modules and data-driven business features.
- Designed and maintained REST APIs and database structures to support core application functionality and integrations.
- Improved system responsiveness and reduced recurring issues through debugging, optimization and performance improvements.
- Collaborated with cross-functional teams in agile development environments to deliver stable and maintainable software solutions.
JavaSpring BootReactTypeScriptPostgreSQLDockerhttps://cbtw.tech/ ↗ - SI02/2019 — 08/2020
Full Stack Engineer · Sigma Computing Solutions
Remote
Developed backend services and frontend components for multiple web-based business applications and internal systems.
- Built database-driven APIs and reporting modules to support operational workflows and client-facing functionality.
- Improved application reliability and usability through bug fixing, testing and incremental performance enhancements.
- Developed responsive dashboards and interfaces compatible across desktop and mobile environments.
- Participated in implementation, deployment, maintenance and feature enhancement activities within agile teams.
Node.jsExpressReactMongoDBREST APIshttp://sigmacomputing.com/ ↗ - ST09/2017 — 12/2018
Frontend Developer · ST Engineering Infa Systems Ltd.
On-site
Developed responsive and accessible frontend interfaces for enterprise web applications using modern frontend technologies.
- Integrated backend APIs and implemented interactive UI functionality across multiple internal systems.
- Improved frontend usability and page performance through UI optimization, debugging and interface refinements.
- Worked closely with backend and QA teams to maintain stable and consistent user experiences across production applications.
- Supported testing, debugging and maintenance activities for web-based systems.
JavaScriptAngularHTML/CSSREST APIshttps://www.infa.com/ ↗
15+ shipped · 7 featured
Things I've shipped
Selected projects across freelance and full-time roles.
Reachfar Zero — AI-Native Middle/Back Office
2026 · AI
AI-native managed middle/back-office platform that auto-resolves reconciliation breaks on a full Cloudflare stack.
Multi-tenant fintech back-office automation built edge-first on Cloudflare. Six AI agents handle position/cash/trade matching, break investigation, swap resets, accruals, financing-rate intelligence and counterparty comms. A per-tenant TenantAgent Durable Object runs embedded SQLite, with a Rules → Workers AI (Llama 3) → Claude API break-resolution pipeline, real-time WebSocket observation layer in Next.js and a client-owned R2/S3 'bucket-first' data model for zero-break clean offboarding.
AI Engineer · agent pipeline, break-resolution logic, edge architecture.
6
AI agents
Zero-DB
edge-native
Multi-tenant
SaaS
Cloudflare WorkersAgents SDKDurable ObjectsClaude APINext.js
JOY — Multimodal AI Assistant Platform
2025 — 2026 · Automation
Production AI assistant with real-time voice, vision and chat - built for a technical-services SaaS.
Built for Hasie Group's technical-assistance domain (energy & multitechnique): OpenAI GPT-4o for text, vision and audio, plus the Realtime API for bidirectional voice over WebSockets. Next.js full-stack with Prisma/PostgreSQL, server-side sessions, conversation persistence and sharing, an admin dashboard with usage analytics, per-user feature toggles and daily token quotas.
Senior AI Engineer · architecture, backend, AI integration, admin tooling.solo full-stack & AI build.
Voice
real-time WebSocket audio
Multimodal
text · vision · audio
SaaS
production
OpenAI GPT-4oRealtime APIVercel AI SDKNext.jsPrisma / PostgreSQL
STXWORX - Decentralized Freelance Escrow Platform
2024 - 2025 · Web3
Bitcoin-secured freelance marketplace with milestone escrow over STX and sBTC.
Built for a Web3 client: Clarity smart contracts on Stacks (Bitcoin L2) handle multi-token escrow, milestone-based fund release and on-chain reputation badges (NFTs). Express + MySQL/Drizzle backend, React 18 + Vite admin and user dashboards, Stacks Connect wallet auth. Property-based fuzz tested with Rendezvous.
Senior Blockchain Engineer · architecture, Clarity contracts, full-stack integration.
STX + sBTC
multi-token escrow
4-stage
milestone release
NFT
on-chain reputation
ClarityStackssBTCReactDrizzle ORM
Bithomp - XRP Ledger Explorer & Toolse
2023-2024 · Web3
Public-facing explorer and services suite for the XRP Ledger, used by traders, NFT collectors and validators.
Worked on the Bithomp frontend: account & transaction explorer, NFT (XLS-20) browsing, XRPL username lookup and validator/network dashboards. Built responsive React/Next.js UIs against Bithomp's XRPL data APIs, with real-time ledger updates, server-side rendering for SEO and i18n across multiple languages.
Frontend Developer · UI engineering, XRPL data integration, performance.
XRPL
ledger explorer
XLS-20
NFT support
SSR
SEO + i18n
ReactNext.jsTypeScriptXRPLREST API
Private IPFS Swarm — Indexer & CDN Acceleratorporting (Sigma)
2025-2026 · Infrastructure
Private IPFS network with a swarm-scoped content indexer and a CDN-like edge accelerator, all stock-Kubo compatible.
Open-source Go toolkit that turns a private IPFS swarm into a small private CDN. A Private Indexer keeps an O(1) CID→providers registry so discovery skips the public DHT; a Swarm Accelerator fronts the swarm with an LRU cache, indexer-guided provider selection, parallel block fetches and hedged retries to cut tail latency. Ships as single-binary services with per-node announcer sidecars, Prometheus metrics and Docker Compose deployment. 100% compatible with stock Kubo — no protocol changes.
Backend / Systems Engineer · architecture, Go services, distributed systems.r · Sigma Computing Solutions.
O(1)
CID discovery
Hedged
low tail latency
Single-binary
Kubo-compatible
GoIPFSlibp2pDockerPrometheus
Vheer — AI Image & Video Creative Studio
2023 — 2024 · AI
Free AI creative studio: text-to-image, text/image-to-video and a full editing toolset behind one dashboard.
AI/backend engineering for a multi-tool creative platform. Built the model orchestration layer routing requests across frontier image and video models (Flux, SeeDream V4, Veo3, LTX Video, Sora 2, MiniMax), plus the inference pipelines and APIs powering text-to-image, image-to-image, text/image-to-video and editing tools (background removal, face detection/blur,color ops, OCR text extraction). Delivered HD, watermark-free output on a free tier with a streamlined, no-expertise-required workflow.
AI / Backend Engineer · model orchestration, inference pipelines, APIs.
6+
AI models integrated
Image + Video
multi-modal gen
HD
no-watermark output
AIGenerative ModelsSora 2FluxInference APIs
Handwritten Character Recognition (CNN)
2026 · AI
Deep-learning OCR for handwritten digits and letters, with a full word-level recognition pipeline.
PyTorch CNN stack that recognises handwritten digits (MNIST) and letters (EMNIST) out of the box, trains alphanumeric models up to 62 classes and segments whole handwritten lines into characters for end-to-end OCR. Ships pre-trained checkpoints, a unified training CLI and two Gradio web apps for single-character and full-text recognition.
ML Engineer · architecture, model training, OCR pipeline, web apps.
99.3%
digit accuracy
94.0%
letter accuracy
OCR
word-level pipeline
PyTorchCNNEMNISTOpenCVGradio
Cross-industry experience
Sectors I've delivered AI and full-stack work into.
- ·
HealthcarePatient-facing tools and back-office automation.
- ·
FinanceData pipelines, dashboards and document processing.
- ·
EducationLearning platforms and AI-assisted content tools.
- ·
SaaSMulti-tenant apps with auth, billing and analytics.
- ·
AutomationWorkflow + RPA-style internal tooling.
- ·
Web3Onchain analytics and LLM-powered summaries.
measured by what shipped, not what was tried
Stack & skills
Four clusters drawn from 8+ years across AI, fullstack, mobile and infra.
✦
AI / ML
Generative AI, NLP, computer vision and the cloud plumbing around them.
- OpenAI APIGPT models, function calling92
- LangChainagents, RAG, tool use88
- PyTorch87
- TensorFlow86
- NLP / LLMschatbots, text processing88
- Vector DBsintelligent search, RAG87
◐
Frontend
Modern React/Next.js and cross-platform mobile.
- TypeScript95
- React / Next.jsApp Router, SSR, RSC94
- Angular87
- React NativeExpo, native modules88
- Flutter86
- Swift / KotliniOS & Android native86
▣
Backend
REST & GraphQL APIs, scalable services, secure auth and payments.
- Node.js / Express92
- Python · FastAPI / Flask90
- Java · Spring Boot87
- Go89
- PHP · Laravel86
- .NET / C#86
◈
Data & Infra
Databases, containers, CI/CD and cloud deployment.
- PostgreSQL92
- MongoDB88
- Docker / Kubernetes88
- AWS / GCP88
- CI/CD · Git90
- Pandas / NumPy87
8+ years · still growing
Tree of Stack
Every leaf is a tool I've shipped with in production.
Web3AI / FrontendSystemsInfra
- 2017Graduated Niigata University · BSc CS
- 2018Frontend Developer · ST Engineering Infa
- 2019Full Stack Engineer · Sigma Computing
- 2021AI & Full Stack · CBTW
- 2022Freelance · Senior AI Engineer
- 2026Leading AI-driven SaaS work in Japan
All technologies
The full toolbox listed on my CV — languages, frameworks, infra.
JavaScriptTypeScriptPythonJavaPHPGoC#RustSwiftKotlinReactReact NativeAngularNext.jsNode.jsExpressFlutterExpo.NETSpring BootLaravelFastAPIFlaskREST APIsGraphQL APIsPyTorchTensorFlowLangChainOpenAI APIPostgreSQLMongoDBPandasNumPyDockerKubernetesAWSGCPCI/CDGit
notes from the build
⌑ RSSWriting
Engineering notes, case studies and the occasional essay.
- ↗
Shipping AI features that actually move metrics
★ featuredNotes on integrating OpenAI and LangChain into production SaaS — what to cache, what to monitor, and how to keep latency below the patience threshold.
2025-11-02·10 minengineering - ↗
20–25% faster: a SaaS backend optimisation playbook
★ featuredThe specific moves — index strategy, N+1 hunting, query rewrites and connection pooling — that took API latency down across a recent freelance build.
2025-08-14·9 mincase-study - ↗
RAG that doesn't lie: grounding LLMs in private business data
Vector DB choices, chunking strategies, eval loops and the guardrails I now ship with every retrieval-augmented system.
2025-05-20·12 mintutorial - ↗
From frontend dev to AI engineer — an 8-year arc
How my career moved from Angular interfaces at ST Engineering, through enterprise fullstack at CBTW, to leading AI-driven SaaS work as a freelancer in Japan.
2025-02-08·7 minessay
Newsletter
Occasional notes on shipping AI products and full-stack engineering from Japan. Short, practical, no fluff.
No tracking. One-click unsubscribe.
Let's build something useful.
If you're working on an AI-driven product, a SaaS that needs smarter features, or a full-stack build that needs an experienced pair of hands — I'd love to hear about it. Short pitch is fine; details can come later.
Pick a time that works
30 min · JST (UTC+9)Loading available times…
Prefer async? Just email donyhorse@gmail.com.
Engagements
Exact scope and rate are agreed before the work starts.
Contract engineering
openProject-based
Solo or embedded with your team. AI integration, full-stack web and backend systems.
AI feature development
openOn request
Chatbots, RAG, NLP and workflow automation built on OpenAI / LangChain / vector DBs.
Code & architecture review
openFlat fee
Async review of an existing codebase + debrief call. AI or fullstack stacks.
Advisory
limitedSelective
For early-stage teams shipping AI-driven products, especially in Japan.
FAQ
- Do you take full-time roles?
- Open to it for the right team — preferably remote-friendly with a clear AI / fullstack mandate. Currently leading a small team in Japan, so timing matters.
- Where are you based / will you travel?
- Nagoya, Japan. Comfortable working fully remote with global teams; can travel within Japan and occasionally internationally for kickoffs.
- What languages do you work in?
- Japanese (native) and English (proficient). Happy to handle bilingual stakeholders.
- What kinds of projects fit best?
- AI-driven SaaS, automation, intelligent search, RAG, and full-stack builds that need both backend depth and a polished frontend.