KKhang
Portrait of Khang

About me

Hi there!I'm Khang

I graduated in Industrial Management, specializing in Supply Chain & Operations Management at Ho Chi Minh City University of Technology. That foundation taught me to look closely at systems, processes, and operational bottlenecks. Today, I build with AI and write down what the process actually teaches — the research, the experiments, and the lessons, with costs and limits in the open. Supply chain is an applied domain for that work, not the limit of what I write about.

My journey

  1. 01
    Foundation

    Supply Chain & Operations

    I graduated in Industrial Management from Ho Chi Minh City University of Technology, specializing in Supply Chain & Operations Management.

    2021-2025
  2. 02
    A change in direction

    From processes to data

    Questions about bottlenecks and decisions led me into analytics in 2023, and then opened the way to AI.

    2023-present
  3. 03
    Current work

    Learning by rolling up my sleeves

    I pick a concrete problem, build a small version, record the assumptions, and open the code to see what holds up.

    present
  4. 04
    Goal

    Learn in public, share what works

    My long-term goal is to keep building, checking what holds up, and sharing what the process teaches. Supply chain remains a domain where I test those ideas against real decisions and KPIs.

    future

My projects

01

Olist Ecommerce Analytics Agent

An ecommerce analytics agent in development for the vibe-coding capstone for the Kaggle 5-Day AI Agents course.

View project ↗

My tech stack

AI providers

AnthropicClaude and agent workflowsOpenAIModels and APIs for prototypesGoogle GeminiModels, ADK, and BigQueryGrokFast model experimentsDeepSeekReasoning and coding modelsGLM / Z.aiModel family from Z.aiQwenOpen-weight modelsKimiModels and agent toolsMiniMaxMultimodal models

Workflows & app building

PythonAnalysis and automationSQLSQLQueries and metric definitionsPower BIData models and dashboardsBigQueryAnalytics warehouseGoogle ADKAgent orchestrationHermes AgentAn autonomous agent for research and coding workflowsClaude CodeA coding agent for development and code verificationCodexA coding agent for building, reviewing, and automationGitVersion control for every projectDockerPackage the runtime environmentKubernetesOrchestrate containerized apps

My thinking framework

I combine McKinsey's seven-step problem solving, Design Thinking, and DevOps to frame the right problem, validate decisions, and improve continuously.

  1. 01

    Define the problem and decision question

    Start with the people involved and the decision to support; clarify the objective, scope, constraints, and success criteria before considering solutions.

  2. 02

    Structure with an issue tree / hypothesis tree

    Break the problem into logical branches using a MECE lens - mutually exclusive, collectively exhaustive - then turn the important branches into testable hypotheses.

  3. 03

    Prioritize by impact and ability to influence

    Focus on the levers that can materially change the outcome and are within reach, rather than spreading the analysis too thin.

  4. 04

    Build an evidence-led workplan

    Define the data, methods, experiments, risks, owners, and evidence threshold required to answer each hypothesis.

  5. 05

    Analyze, test, learn, and prototype

    Start with heuristics and exploratory data analysis, then use data, experiments, and prototypes to learn quickly, test hypotheses, and adjust the direction.

  6. 06

    Synthesize into a governing thought

    Distill the findings into a clear line of reasoning, making the trade-offs and confidence level explicit so the decision is actionable.

  7. 07

    Recommend, deploy, and improve

    Turn the conclusion into an owned action plan; use version control, CI/CD, observability, and feedback loops to ship, measure, and improve continuously.

Contact

Have a question or an idea to compare?