Ho Nghia Khang
AI in SDLC & Chuyển Đổi SốAI Engineer & Applied AI Solutions Architect
AI-Augmented Software Production (SDLC) • Enterprise Digital Transformation • Autonomous Agents • Computer Vision
Executive Summary & Digital Transformation Focus
AI Engineer with over 5 years of hands-on experience pioneering the integration of Artificial Intelligence into the software development lifecycle (SDLC) and driving pragmatic digital transformation (Chuyển đổi số) for businesses and industrial facilities in Vietnam. Expert at architecting autonomous AI agents, RAG knowledge systems, and computer vision pipelines, and embedding them directly into development operations to eliminate repetitive toil, elevate team productivity, and deliver measurable business ROI. Strong engineering foundation from DUT’s elite PFIEV program with 2 registered Vietnamese patent applications (Cục Sở hữu Trí tuệ) and production track records across manufacturing, cloud infrastructure, and edtech.
Featured Applied AI & Digital Transformation Projects
Production Deployments & Case StudiesCodeMath Solver: Autonomous Neural Code Synthesis & Testing Agent
🏆 #1 Leaderboard Rank in 24h- AI in Software Production Showcase: Built an autonomous AI agent demonstrating end-to-end code synthesis, semantic algorithm deduction, automatic unit-test generation, and submission without human oversight.
- Designed a multi-stage prompt chain pipeline: problem decomposition → algorithmic formulation → edge-case synthesis → code generation and formatting.
- Implemented a resilient browser automation layer with exception handling and self-recovery routines, capable of operating unattended for extended multi-hour runs.
- Business / Technical Impact: Automated hundreds of programming challenge solutions, achieving the #1 position on codemath.vn within 24 hours of autonomous launch.
- Open-Source Repository: github.com/LectureHubTeam/codemath-solver
Factory License Plate Recognition & Automated Logistics Checkpoint
Chuyển Đổi Số Nhà Máy • 97% Accuracy- Digital Transformation for Vietnamese Manufacturing: Spearheaded the digitalization of truck entry/exit logging at Vietnamese woodchip factories, completely eliminating manual paper registration and checkpoint bottlenecks.
- Trained and fine-tuned CNN character recognition models robust to harsh environmental conditions: heavy wood dust, direct glare, rain, and angled vehicle positioning.
- Achieved 97% real-world accuracy on private field datasets; packaged into lightweight Docker microservices with a REST API for rapid on-premise edge hardware deployment.
- Publication: Research outcomes peer-reviewed and published in the Proceedings of ATiGB 2022 (ISBN: 978-604-80-7240-7).
- Repository: github.com/KudoKhang/LPR
LectureHub: Competitive Programming Ecosystem & RAG Knowledge Mentor
Chuyển Đổi Số Giáo Dục • Live Production- EdTech Digital Transformation: Built a centralized software platform aggregating programming problems from multiple international judges to support Vietnamese students and engineers in algorithmic upskilling.
- Designed a Retrieval-Augmented Generation (RAG) assistant leveraging vector similarity search with PGVector to provide precise, hallucination-resistant hints and pedagogical walkthroughs.
- Configured full AWS cloud infrastructure (EC2, S3, RDS) with automated CI/CD deployment pipelines ensuring 99.9% uptime.
- Live Website: lecturehub.tech | RAG Chatbot Repo: github.com/LectureHubTeam/lecturehub-chatbox
GPU Infrastructure Virtualization & Resource Slicing Platform
Sep 2025 – Nov 2025- AI Cost Reduction for Vietnamese Tech Teams: Solved the prohibitive cost barrier of high-end GPU hardware for local AI developers by engineering a multi-tenant GPU slicing and dynamic allocation system.
- Automated bare-metal server setup and CUDA runtime configuration using Ansible playbooks, slashing server provisioning time from days to under 15 minutes.
- Developed robust REST APIs managing dynamic resource leasing, containerized tenant isolation, and automated environment teardown.
High-Throughput Asynchronous Microservices for Heavy Compute AI Workloads
Sep 2023 – Jan 2025- Modernizing Legacy Batch Pipelines: Redesigned monolithic media processing services into an event-driven, distributed microservice architecture capable of reliably processing heavy multi-hour video and image generation workloads.
- Engineered non-blocking APIs with immediate response tokens, offloading GPU/CPU tasks to Kafka-coordinated worker pools with automatic retries and dead-letter queues.
- Implemented end-to-end observability (Prometheus, Grafana, Sentry) providing real-time telemetry on worker throughput, GPU VRAM usage, and failure rates.
Tactical Soccer Match Analysis & Dual-Camera Panoramic Video Stitching
Commercial Contract ($1,000+)- Engineered a cross-platform desktop application enabling coaching staff to automate manual camera tracking and perform tactical spatial analytics.
- Built an end-to-end video-stitching pipeline combining two wide-angle 4K camera streams with lens distortion correction, feature matching, and seamless blending.
- Implemented an intelligent electronic pan-tilt-zoom (ePTZ) module that dynamically tracks player centroids in real time.
- Demo Video: youtube.com/watch?v=ZNacjRtL9JQ
Competencies: AI in SDLC & Digital Transformation
AI-Assisted Development, Autonomous Coding Agents, Multi-Turn Prompt Chaining, Tool Use & MCP (Model Context Protocol), Automated Unit/Integration Test Synthesis, AI Code Review & Documentation Generation.
Process Automation, Legacy System Modernization, Distributed Microservices (FastAPI, Kafka, Celery), Non-blocking Asynchronous Architectures, Operational Bottleneck Elimination.
Industrial License Plate Recognition (LPR), Real-Time Video Stitching, Electronic PTZ, Object Detection, Semantic Segmentation (Face/Hair Parsing), OCR, Image Processing, OpenCV, PyTorch.
Docker, AWS (EC2, S3, Lambda, Batch, RDS, VPC), Ansible Infrastructure as Code, GPU Virtualization & Slicing, CI/CD (GitHub Actions, GitLab CI), Linux Administration, Telemetry (Prometheus, Grafana, Sentry).
PostgreSQL, PGVector (Vector Embeddings & Semantic Search), MongoDB, Redis (In-Memory Caching & Deduplication), Knowledge Retrieval Pipelines, Hallucination Reduction.
Rapid PoC validation for digital transformation initiatives: Next.js, ReactJS, Tailwind CSS, Streamlit for internal dashboards, Electron for cross-platform desktop inspection tools.
Professional Work Experience
- Lead the design and rollout of applied AI solutions across computer vision, machine learning models, and automated developer workflows.
- Research and construct autonomous AI agents to automate internal software engineering operations, reduce manual triage, and accelerate sprint cycle velocities.
- Benchmark frontier LLMs, multimodal models, and event-driven agent architectures to identify high-leverage integration opportunities for organizational productivity.
- Modernized cloud infrastructure for AI model training and inference workloads using Ansible automation playbooks.
- Engineered a GPU resource partitioning system that virtualizes expensive enterprise GPUs into smaller isolated rental units, dramatically reducing AI computing costs for engineering teams.
- Built high-reliability REST APIs for programmatic server allocation, user provisioning, and usage metering.
- Developed cloud backend features and proof-of-concept solutions in close collaboration with product owners and frontend developers.
- Architected relational database schemas and queries in PostgreSQL for data integrity and speed.
- Integrated AWS services (S3, Lambda, AWS Batch) to build scalable, serverless, event-driven backend workflows.
- Architected and scaled asynchronous microservices for heavy AI video/image compute workloads using FastAPI, Kafka, MongoDB, and Docker.
- Designed non-blocking API patterns returning immediate processing tickets while background GPU cluster nodes executed prolonged rendering tasks.
- Established robust CI/CD pipelines (GitLab CI, GitHub Actions) covering automated linting, test suites, container builds, zero-downtime releases, and fast rollback.
- Configured production telemetry with Prometheus, Grafana, and Sentry for real-time observability into GPU metrics, service uptime, and error budgets.
- Mentored junior engineers and interns on AI model deployment, containerization, Git workflows, and software reliability best practices.
- Researched and implemented computer vision models for semantic segmentation (face parsing, hair segmentation), object detection, and OCR.
- Reproduced state-of-the-art computer vision papers and benchmarked performance on proprietary datasets.
Patents & Intellectual Property (Cục Sở Hữu Trí Tuệ)
Education
Danang University of Science and Technology (DUT)
2026 – PresentDanang University of Science and Technology (DUT)
2017 – 2022Elite French-Vietnamese engineering curriculum with comprehensive emphasis on control theory, applied mathematics, industrial robotics, and automation engineering.
Community & Tech Upskilling in Vietnam
- Mentored 100+ students in computational thinking, Python fundamentals, and algorithm foundations — nurturing next-generation tech talent for Vietnam’s digital economy.
- Authored customized learning roadmaps demystifying advanced coding and problem-solving concepts into intuitive, hands-on challenges.
- Community Site: scratchzone.tech