# yaml-language-server: $schema=https://raw.githubusercontent.com/rendercv/rendercv/refs/tags/v2.8/schema.json cv: name: Tran Quoc Truong headline: Product Engineer · Applied AI · Backend Systems location: Ho Chi Minh City, Vietnam email: tranquoctruong20@gmail.com website: https://portfolio.truong51972.id.vn/ social_networks: - network: LinkedIn username: truong51972 - network: GitHub username: truong51972 sections: summary: - 'Product Engineer & Applied AI Engineer building agentic applications, RAG platforms, and backend systems, with a focus on production reliability, observability, and evaluation.' - 'Combines formal AI training with backend architecture experience in control-plane design, domain/service boundaries, and asynchronous task processing. Bridges software engineering and AI through trace reconstruction, runtime observability, and automated evaluation gates.' technical_skills: - label: Agentic AI & Orchestration details: LangGraph, Multi-Agent Systems, MCP (Model Context Protocol), Tool Calling, LiteLLM, LangChain, Agentic Workflows, Qdrant (Vector DB), Milvus, Docling, RAG & Hybrid Retrieval - label: LLM Evaluation, Observability & Security details: OpenTelemetry (OTel), Runtime Trace Reconstruction, LLM-as-a-Judge, Deterministic Quality Gates, Prompt Injection Testing, Data Leakage Evaluation, LoRA/QLoRA Fine-tuning, Synthetic Data Generation, Langfuse, PyTorch - label: AI Systems & Backend Architecture details: Python, FastAPI, Django/DRF, Pydantic, SQLAlchemy, Celery, Redis, PostgreSQL, RESTful APIs, Asynchronous Workflows, Idempotent Processing - label: Platform, Infrastructure & Tooling details: Docker, Docker Compose, Nginx, S3-compatible Object Storage (MinIO), Git, CI/CD, Linux experience: - company: FPT Software position: Applied AI Engineer & Automation Tester start_date: 2024-12 end_date: present location: Ho Chi Minh City, Vietnam highlights: - 'IQP (Code Intelligence & Quality Retrieval): Architected backend retrieval for IQP combining semantic vector search, BM25 lexical signals, and Code Graphs, exposing contextual tools to AI clients via Model Context Protocol (MCP) to automate quality analysis across multi-repository codebases.' - 'Agent Assurance (Testing Discipline for AI): Applied software testing discipline to agentic systems by architecting an assurance engine with explicit domain boundaries and async evaluation workflows; implemented evidence-capture paths through LiteLLM and OpenTelemetry-based instrumentation, with idempotent trace persistence.' - 'Agent Assurance (Assurance Lifecycle & Verification): Applied an end-to-end assurance lifecycle spanning Intake, Strategy, Requirements, Test Design & Oracle Definition, Measurement Qualification, Validation, Reporting, and Re-assurance, supported by trace-based behavioral evaluation and regression workflows.' - 'Banking Automation & Quality Operations: Engineered reusable cross-platform test automation and 1,500+ lines of SQL validation across banking and payment systems, cutting manual regression effort by ~70%.' - company: FPT Software Academy position: Full-Stack Developer Intern start_date: 2024-09 end_date: 2024-12 location: Ho Chi Minh City, Vietnam highlights: - Built the core backend for an online learning and assessment platform using Django with role-based access control and automated exam workflows. - Integrated a separate FastAPI computer-vision service for AI-assisted real-time exam proctoring and face detection. - Containerized and deployed the multi-service architecture using Docker Compose, Nginx reverse proxy, and Cloudflare Tunnel. achievements: - name: 2nd Place — IVS Hackathon 2026 date: 2026-08 summary: Awarded 2nd place for independently architecting an end-to-end AI Agent Verification engine (Agent Assurance) featuring runtime evidence capture, trace-based behavioral evaluation, and governed quality gates; subsequently integrated into Omni-Agent. - name: Second Prize — Autonomous Driving Research Paper Competition date: 2024-04 summary: Co-authored research combining YOLOv8 perception, lane segmentation, and PID steering control in Unity simulation, completing the autonomous benchmark in 125.8 seconds with a perfect 100% score. featured_projects: - name: Omni-Agent — Applied AI Platform (Knowledge Builder & Agent Assurance) summary: An Applied AI platform engineered from real operational needs, unifying versioned document RAG, LangGraph multi-agent orchestration, and integrated agent verification within a modular multi-service architecture. highlights: - 'Platform Architecture: Engineered a modular monorepo featuring a Django/DRF control plane for project tenancy, identity, and permissions, backed by decoupled, stateless Celery workers for asynchronous processing workloads.' - 'Knowledge Builder: Built document processing pipeline with Docling layout parsing/OCR, snapshot chunking, Qdrant vector indexing, a Streamable HTTP MCP tool server, and a LangGraph-based multi-agent runtime with citation-grounded RAG.' - 'Agent Assurance (IVS Hackathon 2nd Place): Implemented execution evidence capture through LiteLLM and OpenTelemetry, trace reconstruction pipelines to normalize multi-turn trajectories, and deterministic quality gates.' - 'Reliability & Data Integrity: Designed idempotent evidence ingestion keyed by trace/span identity, with PostgreSQL-backed execution metadata and S3-compatible artifact storage.' - name: APIT — Agent Programmatic Integration Testing summary: AI-assisted API testing framework generating structured test scenarios from API documentation via fine-tuned LLMs. highlights: - Developed a bilingual synthetic dataset generation pipeline producing 1,258 domain-specific API testing samples. - Fine-tuned Qwen2.5-3B with LoRA/QLoRA, achieving macro-F1 of 0.655 (+41.5% relative improvement over Llama-3.2-3B baseline). - Implemented a hybrid evaluation pipeline combining LLM-as-a-Judge, fuzzy matching, and deterministic JSON-schema validation. - name: E-Commerce AI Assistant — Conversational RAG & Search summary: End-to-end shopping assistant combining filtered vector search and multi-agent conversational retrieval. highlights: - Built product semantic search using Milvus, cosine similarity, embeddings, and dynamic category/price filters. - Implemented multi-turn conversational agent with LangGraph and LangChain for context-aware product consultation. - Containerized multi-service architecture with FastAPI, PostgreSQL, Redis, Milvus, and Docker Compose. education: - institution: FPT School of Business & Technology area: Master of Software Engineering in AI start_date: 2026-05 end_date: present location: Ho Chi Minh City, Vietnam - institution: FPT University area: Bachelor of Artificial Intelligence start_date: 2021 end_date: 2025 location: Ho Chi Minh City, Vietnam languages: - label: Vietnamese details: Native - label: English details: Professional working proficiency design: theme: classic page: size: a4 top_margin: 1.35cm bottom_margin: 1.35cm left_margin: 1.35cm right_margin: 1.35cm show_footer: true header: space_below_name: 0.45cm space_below_headline: 0.35cm space_below_connections: 0.3cm section_titles: space_above: 0.38cm space_below: 0.2cm sections: allow_page_break: false space_between_regular_entries: 0.75em space_between_text_based_entries: 0.28em entries: highlights: space_between_items: 0.28em space_above: 0.1cm typography: line_spacing: 0.58em locale: language: english settings: current_date: today render_command: output_folder: public/cv pdf_path: public/cv/Tran_Quoc_Truong_CV.pdf typst_path: public/cv/Tran_Quoc_Truong_CV.typ html_path: public/cv/Tran_Quoc_Truong_CV.html markdown_path: public/cv/Tran_Quoc_Truong_CV.md png_path: public/cv/Tran_Quoc_Truong_CV.png pdf_title: Tran Quoc Truong - CV