Data · AI · Leadership

I build useful
data & AI systems.

And help teams deliver them.

From field engineering across five countries to regional analytics leadership, my work connects hands-on development, business priorities, and people across cultures.

Explore my work Get in touch

**Khoa Pham**Engineer by background.
Builder by practice.

05countries in field engineering

APACanalytics team leadership

MSccontinued study in data & AI

01 / Selected work

Built. Led.
Learned.

Three perspectives on the work:
technical depth, practical impact,
and the curiosity to keep building.

01 / Independent research & development

MedSegAPI

From a thesis to working software.

A lung-segmentation model, an API, deployment, and monitoring. An end-to-end exercise in turning research into something usable.

Explore the repository

Real MedSegAPI output, showing the original chest X-ray, predicted lung mask, green segmentation overlay and model results side by side. — Actual model output. Research software; not clinically validated.

My contributionModel experiments, serving API & monitoring

Built withPyTorch · FastAPI · Docker · GitHub Actions

What I built & learned

I explored how local image features and global context work together in a hybrid CNN–ViT model, then built the serving layer with logging, monitoring, and automated deployment.

The Montgomery + Shenzhen adaptive-hybrid experiment reported 96.65% Dice. Testing dataset shifts and model variants taught me why a benchmark alone cannot establish real-world reliability.

Read the thesis Watch the demo

02 / HEINEKEN · APAC analytics

Making manufacturing
data useful.

I led data engineers and business analysts, translating operational questions into data use cases and aligning technical teams, brewery operations, and global stakeholders.

APAC Data Analytics Manager

Filtration use case180hours

less manual work
per brewery, per year.

Result documented in my CV

UnderstandTurn operational questions into data use cases.

BuildGuide pipelines, architecture and governance.

AlignConnect technical choices to business priorities.

My role & the result

As APAC Data Analytics Lead, then Manager, I guided architecture and governance across analytics for filtration, energy, water, and equipment performance. Industrial IoT pipelines, PostgreSQL, InfluxDB, and Power BI connected sensor data to operational decisions.

Senior stakeholder presentations and regional training helped connect the technical choices to business priorities. The filtration result is documented in my CV.

Experience & results in my CV

03 / Personal proof of concept

Research Radar

From a paper to a short video.

I connected research discovery, assessment, media generation, and delivery into one AI workflow—triggered through Telegram.

Read the build story

Research Radar architecture diagram from Khoa’s LinkedIn post, connecting discovery, assessment, generation, rendering and Telegram delivery.

Inside the workflow

I connected a Telegram trigger to paper discovery, filtering, assessment, script and media generation, video rendering, and delivery. The goal was to learn by making a complete workflow useful.

A related experiment, Finance Brain, explores document extraction and classification with SQLite and a LangGraph ReAct agent. Both are personal proofs of concept.

Read the build story

Actual Airflow project DAG: stage events and songs, load fact and dimension tables, then run data-quality checks.

Also built / Udacity coursework

Data pipelines,
from end to end.

Data modeling, cloud storage, transformation, and orchestration—with checks that make the data usable.

SQL, AWS, Spark, Airflow

See the projects

Also explored: Interview Agent Team hackathon prototype

The team prototype combined LangGraph, Gemma, Qwen, Whisper, and Edge TTS for a voice interview. I took the candidate role in the demo and shared the experiment.

See the hackathon story

More projects on GitHub

02 / How I lead