AI Engineer · RAG · Data SystemsPakon Poomson

AI products and data systems built for real workflows.

AI Engineer building GenAI workflows, RAG systems, multimodal apps, backend APIs, and data platforms for real operational use cases.

View Projects Resume GitHub LinkedIn

Customer support RAG triage dashboard and workflow concept.

Selected engineering evidence

Customer Support RAG Triage Agent

Case study

Current role: AI Engineer at Seagate Technology
Core focus: RAG, agentic workflows, multimodal AI
Delivery: FastAPI, React, data systems, deployment

Selected Work

Three evidence-led case studies across agentic RAG, multimodal product engineering, and time-series ML evaluation.

Case studyPython + FastAPI + LangGraph

Customer Support RAG Triage Agent

Support teams need consistent ticket triage, relevant evidence, and grounded draft responses without relying on a generic chatbot.

What I built

System: A retrieval-grounded support triage system with observable agent steps, safe fallback behavior, and measurable evaluation.
Implementation: Fixed support workflow with typed state instead of an open-ended chatbot loop.

What it proves

Skills: RAG architecture, LangGraph orchestration, vector retrieval, provider routing, evaluation design, FastAPI, React, Docker, and deployment troubleshooting.
Core stack: Python, FastAPI, LangGraph, Qdrant

RAGLangGraphQdrantEvaluation

Details GitHub

Demo availableNext.js 16 + React 19 + TypeScript

Receipt AI Expense Tracker

Manual expense entry is slow and error-prone when receipts mix Thai and English text, inconsistent layouts, and Buddhist Era dates.

Receipt expense tracker dashboard with totals, categories, and receipt history.

What I built

System: A multimodal receipt-to-expense workflow with safe provider routing, human review, local persistence, and analytics.
Implementation: Image-capability filtering prevents receipt images from being sent to text-only providers.

What it proves

Skills: Multimodal AI product design, Thai/English normalization, schema enforcement, local-first architecture, Next.js, IndexedDB, and testable provider fallback.
Core stack: Next.js 16, React 19, TypeScript, Gemini

Multimodal AILocal-firstThai/EnglishAnalytics

Details GitHub Demo

Case studyPython + pandas + statsmodels

Climate CO2 Forecasting ML

Atmospheric CO2 forecasting needs leakage-safe time-series evaluation rather than a visually impressive model trained on future information.

CO2 Forecast Lab dashboard showing model comparison and atmospheric trend.

What I built

System: An end-to-end forecasting and anomaly-analysis system with reproducible data preparation, model comparison, API serving, and dashboard evidence.
Implementation: Chronological train, validation, and test splits prevent future leakage.

What it proves

Skills: Time-series methodology, leakage prevention, honest model evaluation, statistical forecasting, PyTorch experimentation, FastAPI, React, and reproducibility.
Core stack: Python, pandas, statsmodels, scikit-learn

ForecastingTime seriesEvaluationAnomaly detection

Details GitHub

Explore all 6 case studiesAI engineering, data systems, and applied ML evidence

About

Practical engineering across models, APIs, data, interfaces, and deployment.

I am an AI Engineer at Seagate Technology building practical GenAI automation, internal engineering tools, and backend data workflows. My public projects demonstrate production-style thinking across retrieval, multimodal extraction, document analysis, forecasting, and explainable ML.

I use personal projects to show the parts that are often missing from AI demos: input validation, provider routing, safe fallback, measurable evaluation, honest limitations, and interfaces people can actually use.

How I work

1Start with the workflow and failure modes, not the model demo.
2Keep provider boundaries, validation, and fallback behavior explicit.
3Treat deployment, evaluation, and clear limitations as part of the product.

Based in Nakhon Ratchasima, Thailand. Targeting AI Engineer, GenAI Engineer, Data Engineer roles.

Capabilities

The stack behind the public case studies, grouped by the work it enables.

AI & Retrieval

Python
LangGraph
RAG
Qdrant
Embeddings
Gemini
Groq
Cerebras
Multimodal extraction
Structured outputs
Offline evaluation

Frontend

Next.js
React
TypeScript
Tailwind CSS
Recharts
Responsive UI

Backend & APIs

FastAPI
Pydantic
REST APIs
Node.js
SQLAlchemy
Provider routing
C#
Java

Data & Infrastructure

PostgreSQL
pgvector
IndexedDB
Apache Kafka
Apache Flink
VictoriaMetrics
Time-series data
Data normalization
Ingestion validation

Delivery & Operations

Git
GitHub
Docker
Vercel
Render
GitHub Actions

Experience

AI engineering, full-stack delivery, automation, and data-infrastructure work.

Jan 2026 - Present

AI Engineer

Seagate Technology

Builds practical GenAI automation and internal engineering workflow tooling while supporting backend and data-infrastructure work.

Responsibilities

Develop multi-agent workflows for requirements analysis, debugging, testing, and pull-request preparation.
Build and maintain internal C# engineering software.
Support time-series storage and Kafka-to-TSDB ingestion validation.

Achievements

Applied agentic AI patterns to repeatable engineering workflows.
Contributed to VictoriaMetrics TSDB setup and ingestion verification.
Connected AI application work with operational backend and data concerns.

Technologies

GenAI agents
C#
VictoriaMetrics
Apache Kafka
Apache Flink
Java

Sep 2023 - Jan 2026

Full-stack Developer

WANG CORPORATION CO., LTD.

Delivered full-stack applications and internal automation tools spanning computer vision, data extraction, civic-safety systems, multilingual meeting workflows, and operational web applications.

Responsibilities

Designed React and Next.js interfaces backed by APIs and relational data.
Built AI-assisted extraction, summarization, and computer-vision workflows.
Delivered internal tools for queueing, maintenance requests, and reporting.

Achievements

Built reusable full-stack and automation patterns across multiple internal products.
Converted unstructured inputs into structured datasets and operational dashboards.
Shipped Dockerized services and data-backed applications for remote stakeholders.

Technologies

React
Next.js
FastAPI
PostgreSQL
Directus
Supabase
Docker

How I Can Contribute

Practical delivery areas supported by the public projects and current engineering experience.

From uncertain inputs to a deployable workflow

I can contribute across AI-enabled applications, retrieval workflows, structured extraction, backend APIs, data systems, and the React interfaces that make those systems understandable.

Open to

AI Engineer roles
Full-stack Developer roles
Software Engineer roles
Freelance AI automation projects
MVP builds for AI-enabled products

AI application prototypes

Build working AI product prototypes that connect user input, model calls, structured output, and a usable web interface.

AI Resume MatcherReceipt AI Expense TrackerCustomer Support RAG Triage Agent

RAG and document-grounded chat

Create retrieval workflows for documents, embeddings, vector search, chat history, and streamed assistant responses.

Customer Support RAG Triage AgentThai Procurement Intelligence

Multimodal extraction workflows

Turn images, receipts, screenshots, product photos, and PDFs into normalized records that can be searched or analyzed.

Receipt AI Expense TrackerAI Resume Matcher

Full-stack web applications

Ship recruiter-ready web apps with Next.js, React, TypeScript, API routes, backend services, and deployment setup.

Next.jsReactTypeScriptFastAPIVercelSupabase

Backend APIs and data systems

Design API flows for ingestion, validation, search, analytics, persistence, and AI service integration.

FastAPIPostgreSQLpgvectorSQLAlchemySupabase

Project evidence is linked to public GitHub repositories and deployments. Employment claims are limited to the confirmed resume and profile material used for this portfolio.

Email Me LinkedIn

Let’s Connect

Direct contact, public engineering evidence, and an updated resume.

Interested in practical AI work? Let’s talk.

Email is the fastest way to reach me. GitHub contains the implementation evidence, LinkedIn covers current experience, and the resume provides a concise career summary.

Email me Resume GitHub LinkedIn