CV
Yan Paing Oo
AI-Native Full-Stack & Cloud Engineer
+6588102190 · yanpaingoo.dev · LinkedIn · GitHub · [email protected] · WhatsApp
AI-native Full-Stack & Cloud Engineer with experience building Go, TypeScript, and cloud-native systems across AI products and regulated digital-market infrastructure. Delivers production services end to end, from API and data design through React interfaces, AWS infrastructure, CI/CD, and operational reliability.
Skills
| Area | Technologies |
|---|---|
| AI / LLM | AI-native development, coding-agent workflows, multi-agent systems, tool-calling agents, RAG, context engineering, prompt design, LLM benchmarking, embeddings, vector search, LangChain, LangGraph, streaming conversations |
| Models | Gemini, ChatGPT, Mistral, Ollama, CloudWeGo Eino, Google GenAI SDK |
| Cloud / DevOps | AWS (EKS, ECS, VPC, IAM, RDS, Secrets Manager, CloudWatch), GCP, Docker, Kubernetes, Terraform, Ansible, GitHub Actions, GitLab CI, Nginx |
| Backend | Go, Python, Node.js, TypeScript, NestJS, PostgreSQL, MongoDB, REST API, WebSockets |
| Frontend | React, Next.js, Vite, Shadcn UI, TanStack |
Experience
AI-Native Full-Stack & Cloud Engineer (Consultant), Jul 2026 - Present
Hydra X, Singapore
- Joined as the sole engineer on a greenfield initiative, partnering with a project manager to turn operational requirements into production systems.
- Shipped transaction-monitoring and reconciliation tools within the first two months, giving operations real-time visibility into discrepancies and alerts for follow-up.
- Built Go services and a Hyperledger Besu adapter to securely connect internal systems with blockchain nodes and support transaction workflows.
- Delivered full-stack internal products, pairing Go backend services with React operational dashboards for engineering and business teams.
- Operate containerized services on AWS EKS and ECS, and extend Terraform modules to improve repeatability across infrastructure and deployments.
- Apply AI-native, specification-first engineering practices, using coding agents for scoped implementation, test generation, QA support, and documentation while retaining human ownership of architecture, security, and review.
AI Engineer & DevOps, Aug 2025 - Jun 2026
Pluggi
- Designed multi-agent LLM architecture with domain-specialized sub-agents and tool-calling. Achieved 100% tool accuracy and 20% token reduction on a 55-case benchmark I designed.
- Engineered RAG evaluation pipeline using RAGAS. Applied BM25 hybrid ranking with vector search to improve retrieval, reducing answer irrelevancy scores by measurable margins.
- Formulated structured response format system (SUMMARY + DATA + GUIDANCE) for agent outputs. Improved agent quality score from 4.7 to 5.0/5.
- Led prompt engineering strategy across 4 domain areas. Designed few-shot examples, chain-of-thought templates, and context injection patterns per domain.
- Defined LLM evaluation methodology: offline benchmark suite plus production sampling, for pre-deploy and continuous quality.
Software Engineer & DevOps (Part-time → Full-time), Feb 2025 - Jul 2025
Pluggi
- Production chatbot struggled at 10K+ U.S. user scale. Architected AWS load balancing with Nginx reverse proxy on reserved EC2 and spot on-demand nodes. Complaints reduced 50%, uptime improved 97% → 99% at 0.57% cost increase.
- Platform had no backend infrastructure. Led architecture from zero with NestJS. System now handles 500+ daily active conversations.
- Built in-house bot-system builder with React Flow visual UI. Workflow deployment speed increased 3x for non-technical users.
- Co-authored state management system using XState with custom extensions. Stabilized flow execution. Mentored 2 junior developers on backend and RAG systems.
Associate Software Engineer, Feb 2024 - Jan 2025
Brillar Pte. Ltd.
- Chatbot responses were slow and limited to single model/channel. Delivered LangChain-powered solutions integrating ChatGPT, Gemini, and Mistral with real-time streaming across 3 channels. Response latency reduced 20%.
- Gen-AI team’s manual deployments were slow and error-prone. Implemented CI/CD pipelines with GitHub Actions and Docker. Deployment cycle reduced 30%.
- Mobile testing across 50+ devices required manual coordination. Built real-time testing platform integrating Appium (Android & iOS), Selenium, and Tosca. Enabled automated parallel testing across 50+ devices.
Freelance Full-Stack Developer & Infrastructure Consultant, May 2023 - Feb 2024
- Client needed a custom financial tracking system. Engineered full-stack Car Ledger Software with inventory tracking and financial logging. Delivered to production.
- License registration required a complex client-specific workflow. Designed and built the entire frontend UI from scratch: a custom multi-stage form with localization and custom font support, wired end-to-end to backend APIs. Demonstrated full-stack delivery across UI and services.
- Ledger systems needed to handle high transaction volumes. Built high-performance Go microservices for data-heavy processing. System handled 100K+ daily transactions.
- Small business clients had no cloud infrastructure or backup strategy. Managed end-to-end deployment with Nginx, SSL, automated backups. Clients gained reliable, secured production environments.
Automation & Performance Engineer, Aug 2022 - May 2023
Brillar Pte. Ltd.
- A Singapore payments network relied on slow manual E2E testing that bottlenecked releases. Executed 150+ automated test scripts using Tricentis Tosca. Test coverage increased 35%, manual testing reduced 40%.
- Hong Kong banking system needed throughput and reliability validation before production. Crafted performance/load test plans with JMeter and led API, load, and stress testing. API throughput improved 30% with 99.9% reliability under production traffic.
- CityNexus, CDL’s smart-lifestyle experience app, needed cross-platform quality coverage before launch. Performed end-to-end mobile and web testing across the app’s features. Ensured a reliable launch experience for returning City users.
Education
University of Information Technology
Bachelor of Computer Science in Knowledge Engineering (AI, ML & Computer Vision)
Certifications
- GCP Professional Cloud Architect
- Automation Specialist Level 1 & Level 2
- NeoLoad Specialist
- ITPEC Common Examination
Projects
Skills & Competency Graph with LLM Chat, 2026
- Natural-language query system over Singapore’s SkillsFuture data, modeled as a typed in-memory knowledge graph (NetworkX, 120 nodes / 237 edges). FastAPI REST API exposes learning-path generation (prerequisite-chain BFS + topological sort), multi-factor gap analysis, and cross-economy transferable-skill lookup.
- Two-call LLM pipeline (intent parsing → graph query → NL answer) on LiteLLM, provider-swappable via one env var (Ollama/Qwen3, Gemini, OpenAI, Anthropic). 3-layer fuzzy resolver (ID → acronym → token-set) maps free text to graph nodes reliably on a small local model.
- React + TypeScript (Vite) frontend with a chat view returning structured answers plus a focused subgraph, and an interactive force-graph explorer. ~33 pytest unit tests + 15/15 end-to-end LLM eval suite.
- Tech: Python, FastAPI, Pydantic, NetworkX, LiteLLM, rapidfuzz, pytest, React, TypeScript, Vite, TanStack Query, react-force-graph, shadcn/ui.