SEK·Experience ↗

JUN — SEP 2025 / CASABLANCA

AKWA GROUP

Data Scientist Intern

Built an end-to-end machine-learning pipeline to detect and classify missing transaction records, from preprocessing and feature engineering through model comparison, evaluation, and the path into an ERP data workflow.

  • Data quality
  • Classification
  • Feature engineering
  • ERP workflow

01 / CONTEXT

Missing records are an operational data problem

The work focused on recognizing absent transaction records and assigning useful classifications so anomalies could move through an existing enterprise process instead of remaining an unstructured investigation.

02 / CONTRIBUTION

Owned the machine-learning workflow

  • Prepared and transformed transaction data for modeling.
  • Designed features and compared supervised classification approaches.
  • Evaluated models and considered how outputs would connect to an ERP data workflow.

03 / METHODS

From data preparation to operational output

High-level workflow
01

Inspect and preprocess source transaction records.

02

Engineer signals for missing-record detection and classification.

03

Train and compare candidate supervised models.

04

Package predictions for an ERP-oriented review path.

The diagram describes responsibilities at a public level; it is not an internal architecture or data-flow export.

04 / OUTCOME

A complete applied-ML delivery cycle

The internship demonstrated how preprocessing, model selection, evaluation, and operational integration must be treated as one system. No numerical accuracy, savings, or production-impact claim is published because supporting public evidence is unavailable.

05 / DISCLOSURE BOUNDARIES

What is intentionally not published

  • No internal dataset, feature definition, transaction sample, or ERP configuration is disclosed.
  • No confidential model metric or business-impact estimate is presented.
  • No public source repository is claimed for employer work.