I am an AI Engineer at Seagate Technology focused on practical GenAI automation, internal engineering workflow tooling, and backend/data infrastructure. My current work includes multi-agent AI workflows for requirements analysis, debugging, testing, and PR preparation, plus C# internal software, VictoriaMetrics TSDB setup, and Kafka-to-TSDB ingestion validation with Apache Flink/Java.
Previously, I built full-stack applications and internal automation tools with React, Next.js, FastAPI, PostgreSQL, Directus, Supabase, and Docker. I focus on turning AI prototypes into usable software: reliable APIs, clean data flows, practical UX, and deployment-ready architecture.
| Priority | Target roles | Evidence in this profile |
|---|---|---|
| 1 | AI Engineer | RAG apps, LLM agents, multimodal extraction, AI automation, evaluation-minded delivery |
| 2 | GenAI Engineer / Data Engineer | LangGraph, Qdrant, Gemini/Groq/Cerebras, pgvector, PostgreSQL, Kafka/Flink validation, VictoriaMetrics |
| 3 | Data Science / Data Analyst / Full-Stack / Frontend | ML projects, analytics dashboards, React/Next.js, FastAPI, SQL, dashboards, API integrations |
| Area | Tools and Strengths |
|---|---|
| AI / ML | Python, TensorFlow, PyTorch, scikit-learn, LLMs, RAG, LangGraph, Qdrant, Gemini, Groq, Cerebras, Ollama |
| Data / Infrastructure | Apache Kafka, Apache Flink, VictoriaMetrics, PostgreSQL, pgvector, Redis, SQL, time-series data, ingestion validation |
| Software Engineering | TypeScript, Next.js, React, Node.js, FastAPI, C#, Java, Docker, CI/CD, Git/GitLab, REST APIs |
| Applied AI | Agentic workflows, computer vision, OCR/entity extraction, multilingual summarization, document retrieval, structured AI outputs |
- AIAT Super AI Engineer Season 6: Foundation AI (Theory) - Artificial Intelligence Association of Thailand, 2026
- Anthropic Academy certificates across Claude API, Claude Code, Model Context Protocol, subagents, agent skills, and AI Fluency
- Google Cloud AI/ML skill badges across Vertex AI, Gemini, Imagen, Multimodal RAG, BigQuery ML, Document AI, and ML APIs
- AIS Academy Prompt Engineering & Agentic AI
| Project | What I built | What it proves |
|---|---|---|
| Customer Support RAG Triage Agent (demo) | LangGraph workflow for intent classification, urgency detection, Qdrant retrieval, grounded response generation, provider fallback, caching, and offline evaluation | RAG architecture, agentic workflows, retrieval evaluation, FastAPI, React, Docker |
| Receipt AI Expense Tracker (demo) | Local-first multimodal product that parses Thai/English receipts, validates structured output, normalizes Buddhist Era dates, and stores reviewed expenses in IndexedDB | Multimodal AI, provider routing, Zod contracts, local-first design, Next.js analytics |
| AI Resume Matcher (demo) | PDF-to-job analysis platform with strict validation, multi-provider routing, skill-gap guidance, interview preparation, deterministic mock mode, and Vercel deployment | Production-minded AI APIs, Pydantic schemas, safe rendering, testing, React/FastAPI delivery |
| Project | What I built | What it proves |
|---|---|---|
| Climate CO2 Forecasting ML | Reproducible time-series pipeline comparing baseline, statistical, scikit-learn, and PyTorch models with shared evaluation and anomaly analysis | Leakage-safe forecasting, chronological validation, model comparison, FastAPI, React |
| Explainable Cancer Diagnosis ML (frontend showcase) | Breast-cancer classification benchmark with safety-relevant metrics, SHAP explanations, strict inference contracts, and an evidence dashboard | Explainable ML, error analysis, model governance, scikit-learn, PyTorch, SHAP |
| Thai Procurement Intelligence (demo) | Thai/English procurement search and analytics platform with ingestion, normalization, hybrid retrieval, dashboards, and evidence-backed Q&A | Data engineering, public-data systems, PostgreSQL/pgvector, FastAPI, Next.js, LLM integration |
- Building GenAI workflow automation for requirements analysis, debugging, testing, and PR preparation.
- Shipping RAG, multimodal AI, and agentic applications with production-minded APIs and data flows.
- Strengthening backend/data infrastructure with time-series storage, Kafka/Flink validation, and PostgreSQL/pgvector.
- Hardening AI applications with better evaluation, observability, testing, and deployment workflows.
- Email: pakon.poomson@gmail.com
- LinkedIn: linkedin.com/in/pakon-poomson
- Portfolio: pakon-portfolio.vercel.app