Portfolio

Location: Yokohama & Tokyo, Japan

SYS.ONLINE

Hello, I'mBojiang Zhang.

Open to:
Remote roles Japan-based roles Australia-based roles
Role:
Clinical Data Engineer · Oncology
Skill:
CDISC SDTM/ADaM · EDC · ETL pipelines

Specializing in Clinical Data Engineering.

Oncology data, CDISC SDTM and ADaM.

Get in touchDownload CV

01_What_I_Do

I turn raw oncology study data into CDISC-conformant SDTM and ADaM datasets, and I build the tools that make that work faster and easier to trust. Below is what that looks like, on made-up data.

Raw EDC export

症例番号 Subject No.
001
性別 Sex
男 Male
生年月日 Date of birth
1958/03/04
有害事象 Adverse event
悪心 Nausea
CTCAE Grade
2
発現日 Onset date
2024/05/12

DM Demographics

STUDYIDDOMAINUSUBJIDSEXBRTHDTC
ONC01DMONC01-001M1958-03-04

AE Adverse Events

USUBJIDAESEQAETERMAEDECODAETOXGRAESTDTC
ONC01-0011NauseaNausea22024-05-12
A synthetic example of the daily work: one row of a Japanese EDC export, mapped into the SDTM DM and AE domains. Hover or focus a raw field to see where it goes.
oncology studies whose data I handled on my own
7
patients across those studies
10,000+
less delivery time after I wrote the team's SOPs
~30%
fastest protocol-to-deliverable cycle
2 months

02_Selected_Works

Open-source projects from my SDTM work. The company data stays private; the methods don't need to.

Knowledge & Spec

SDTM PediaThe SDTM Implementation Guide, rebuilt as a knowledge base that people and LLMs can both search.PythonMarkdownLLM toolsSDTMIG v3.4, SDTM v2.0 and CDISC terminology source files turned into 293 structured Markdown files63 domains, 1,005 codelists and 37,939 CDISC terms, fully indexedWorks with Claude Projects or Cursor directly, with no vector databaseAlongside that Markdown core, an optional RAG and knowledge-graph service with a streaming web chat is in developmentSDTM Spec ForgeGenerates SDTM datasets from a mapping config, then writes their dataset-specification workbooks against SDTMIG v3.4 and CDISC terminology.PythonRustSDTMIG v3.4A Python pipeline generates the datasets from the mapping configA Rust command-line tool then reads those datasets and writes the specification workbooks

Pipeline & Production

SDTM Mapping SystemConfig-driven ETL from raw study exports to SDTM datasets and an M5 submission package.PythonpandasMySQLCDISC SDTMA 7-step automated pipeline, driven by one Excel mapping workbookStudy teams change mapping rules in Excel, with no Python coding

Tooling & Audit

MedAudit Diff WatcherA desktop and command-line tool that watches versioned data folders and records row- and field-level CSV changes in an SQLite audit trail.PythonPySide6SQLiteDataForge StudioA Windows desktop toolbox for SDTM data prep (conversion, cleaning, masking) for analysts who don't write Python.PythonPySide6

Now building

Private for now; these follow on from the projects above.

03_Experience

  1. Sep 2023 – now

    Clinical Data Specialist, Oncology

    iTms K.K., Yokohama/Tokyo

    • Delivered data management for 7 oncology studies (600–4,300 patients each) using data from the National Cancer Center Japan, from protocol to SDTM/ADaM datasets and analysis outputs.
    • Designed study databases and specs, programmed edit checks agreed with hospital clinicians, and produced baseline, AE, lab, OS and PFS outputs.
    • Wrote the team's first checklists and SOPs, and trained 5+ new data staff.

  2. Sep 2022 – Aug 2023

    EDC Developer, Clinical Trials

    EPS Corporation (CRO)

    • Built and maintained EDC databases for oncology trials: eCRFs, edit checks, UAT and user training.

  3. Dec 2017 – Aug 2022

    Test / QA Engineer

    DAC-ENGINEERING K.K.

    • Calibrated and tested precision optical and inspection equipment.

Toolkit

04_Lab

Things I build outside clinical data, each with a live demo.

05_Connect

The quickest way to reach me is email. I reply in English, Japanese or Chinese.

https://github.com/hakupaoLinkedInCV (PDF)[email protected]
Scroll to more