Hi, I'm

Aphiwish Aphisaksiri

Neal·Software Engineer

📍 Bangkok, Thailand

Software engineer shipping agentic AI systems end-to-end — RAG pipelines, LLM tooling, and the production infrastructure behind them.

GitHubLinkedInDownload Resume

About Me

Biomedical Engineer turned AI Software Engineer. I spent four years at Mahidol University building signal-processing systems and competing on the international stage with BCI research — which taught me to care about systems that work under pressure, not just in demos. Today I design and ship agentic AI pipelines, RAG architectures, and the full-stack infrastructure that makes them actually usable.

🏆Grand Prize · 100+ Global Teams8th Delta International Contest🌍International BCI CompetitionCYBATHLON 2024🎓Engineering, Mahidol UniversityFirst Class Honors · Ranked #1🚀Projects Shipped5+

Experience

  • Designed and built a Summary Agent microservice in Go for Re-KYC automation, implementing a custom PII-filtering layer to ensure merchant data privacy and regulatory compliance.
  • Integrated three AI agents (OCR, website verification, and summary generation) into a legacy Ruby on Rails platform, modernizing manual risk-assessment workflows for analysts.
  • Owned the end-to-end service lifecycle from architecture and database migrations to GitOps deployment, ensuring high quality via table-driven tests and automated CI/CD pipelines.
  • Collaborated cross-functionally with Legal, Product, and Compliance teams to translate complex regulatory requirements into technical AI solutions during biweekly steering meetings.
Go
Ruby on Rails
AI Agents
GitOps
Microservices

Projects

A full-stack project management platform with full CRUD for projects, tasks, and notes — built on Next.js App Router with React Server Components, Server Actions, and NextAuth v4 authentication. A separate async FastAPI service owns the AI pipeline: RAG-powered agentic chatbot with pgvector semantic search, a tool-calling loop up to 5 iterations, native thinking mode, and real-time multiplexed streaming surfacing generation metrics. The 4-service stack runs in Docker Compose with NVIDIA GPU passthrough, CI via GitHub Actions, and is deployed to projecthub.aphiwish.com through a Cloudflare-secured tunnel.

⚡185 tok/s inference (Gemma 4)⏱1.85s avg response time🔧5-iteration agentic tool loop
Next.js
FastAPI
PostgreSQL
pgvector
Docker
Ollama
TypeScript
Python
GitHubLive Demo

Skills

🤖AI / ML

Python
OllamaOllama
RAG Pipelines
pgvector
Scikit-Learn
PyTorch
Embeddings
Prompt Engineering

🎨Frontend

Next.js
React
TypeScript
Tailwind CSS
HTML
CSS

⚙️Backend

FastAPI
Node.js
PostgreSQL
Prisma ORM
REST APIs
WebSockets

🛠DevOps & Tools

Docker
Docker Compose
GitHub Actions
Git
Linux
WSL2

Contact

Let's Connect

Open to opportunities, collaborations, or just a chat.

[email protected]github.com/Aphiwish-Aphisaksirilinkedin.com/in/aphiwish-aphisaksiri