Skip to content
View vnlebaoduy's full-sized avatar
🏦
A waiter at the PVcomBank
🏦
A waiter at the PVcomBank

Block or report vnlebaoduy

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
vnlebaoduy/README.md

About

  • DevOps engineer with 3+ years in multi-cloud (AWS, GCP) and on-premise infrastructure, preceded by 5+ years as a developer and R&D engineer.
  • Currently a Center Specialist at PVcomBank, leading the rollout of AI-DLC (AI-Driven Development Life Cycle) guidelines across the entire IT Division — defining how AI coding assistants are used from requirements through release, with guardrails for banking security and compliance.
  • I work across the whole stack, which is what makes the architectural calls possible: event-driven serverless pipelines, container orchestration, IaC, production ML/AI serving, and Python/Java/React application code.
  • Based in Ha Noi, Vietnam 🇻🇳 · English (professional working proficiency).

Certifications

CKS CKA
AWS SAA ITIL Foundation

Stack

cloud, containers, IaC and CI/CD
data, messaging and observability
languages, frameworks and ML

No icon exists for these, so they are listed in text: Proxmox, VMware, KVM, Oracle Database, PostGIS/pgvector, vLLM, YOLO/SAHI, Wazuh, Falco, Trivy, Graylog.


Selected engineering work

Most of this is internal to employers rather than public repositories, so it is described here instead of linked.

Project Scope Outcome
AI-DLC adoption · PVcomBank SDLC governance for AI coding assistants in a regulated bank: guardrails, data-classification rules, human review checkpoints, playbooks and training. Adopted across the entire IT Division.
Ship detection pipeline · GCP Event-driven hybrid compute — spot GCE MIG for I/O-bound 16→8-bit conversion, Cloud Run on L4 GPU for YOLO OBB + SAHI inference, wired through GCS notifications, Pub/Sub and Eventarc with a DLQ. Terraform-managed. 12K+ files/day · 3× cost reduction (~$5,400 → ~$1,800/month) · 35–45s warm latency · ~75% storage compression.
GeoGPT On-premise Qwen3-14B served with vLLM, RAG on PostgreSQL + pgvector (no separate vector DB), natural-language queries over AIS vessel tracking and satellite imagery, plus offline geocoding for air-gapped operation. Full data sovereignty, zero per-token cost, sub-second retrieval.
Space FM Multi-tenant STAC platform for petabyte-scale imagery: pgstac partitioning, ARQ ingestion workers, TiTiler COG tiles, Next.js + MapLibre frontend, RBAC with 60+ permissions and short-lived STS credentials. 100M+ STAC items · <500ms p95 search on 10M · <200ms tile generation · 315+ backend tests.
Datacenter buildout · Vegastar End-to-end design to Uptime Institute Tier II principles: N+1 UPS with diesel generator and ATS, CO2 suppression, precision cooling, 25GbE/SFP28 fabric and 20Gbps SAN, plus vendor and BOM ownership. 53+ VMs provisioned via Terraform/Ansible across Proxmox/KVM and VMware.
Geospatial training platform · AWS Open edX on Auto Scaling, ELB and CloudFront, provisioned with Terraform, hardened with WAF and encryption, monitored with CloudWatch. 3,000+ concurrent users.

Also: SOC build to Vietnam Level 3 information-security standard (Graylog, SIEM/XDR/SOAR, Golang automation) · Traffic Police command centre (Oracle/RMAN, 50M+ records) · AIS data synchronisation on Kafka.

Awards at Vegastar: Best Product of the Year (2025, 2018), Employee of the Year (2023), Second Prize in the Creative Product Contest (2019).


Open source

Repository What it is
pi-kiro-connect Kiro (AWS CodeWhisperer/Q) provider for pi and Oh My Pi — kiro-cli-compatible login, live model catalog, credit usage.
KiroCrew A persistent workspace for development work that self-improves and continues beyond one session.
aidlc-workflows AI-DLC adaptive workflow steering rules for AI coding agents.
aidlc-dashboard-v2 Read-only web dashboard for an AI-DLC run: status, progress, blockers, timing and usage.
fastapi-keycloak-boilerplate FastAPI boilerplate with user management and authentication via Keycloak.
airflow-minio-postgres-fastapi Data-store stack with GitHub Actions CI/CD doing SSH deploy to remote Docker Compose.

GitHub activity

3D contribution calendar profile summary repos per language most committed language stats productive time
contribution streak

Open to work on platform engineering, multi-cloud architecture and AI-assisted delivery workflows — say hello.

Pinned Loading

  1. KiroCrew KiroCrew Public

    Forked from kirodotdev/KiroCrew

    A persistent workspace for development work that self-improves and continues beyond one session.

    Python

  2. aidlc-workflows aidlc-workflows Public

    Forked from awslabs/aidlc-workflows

    AI-Driven Life Cycle (AI-DLC) adaptive workflow steering rules for AI coding agents

    TypeScript

  3. aidlc-dashboard-v2 aidlc-dashboard-v2 Public

    Read-only web dashboard for an AI-DLC v2 run: status, progress, blockers, timing and usage.

    TypeScript