Naef
Aladwan

Configured a standardized Python workspace with virtual environments, linting, and Git workflows to ensure consistent results across team members.

Contributed to a shared codebase using feature-branch workflows, conducting code reviews and resolving merge conflicts to maintain a clean history.

Built an end-to-end data loading and preprocessing pipeline with validation checks at each stage, producing clean, reproducible outputs for downstream modeling.

Implemented a complete train-evaluate cycle in PyTorch — from data loaders and tensor operations through forward/backward passes to metric reporting.

Designed an analytical query pack and KPI brief against a multi-table PostgreSQL database, translating business questions into performant SQL.

Produced a descriptive analytics report — extracted data from PostgreSQL, defined and statistically validated KPIs, created publication-quality visualizations, and wrote an executive summary.

Built a complete train-evaluate loop in scikit-learn using the Petra Telecom churn dataset, comparing models and interpreting evaluation metrics to inform retention strategies.

Built a Named Entity Recognition pipeline on real climate journalism articles, preprocessing raw text and extracting structured entities using spaCy.

Built a pipeline that accepts natural language queries, retrieves semantically similar documents using DistilBERT embeddings, and enriches results with named entities.

Implemented the retrieval backbone of a RAG system on a technical Q&A corpus — ingesting documents, generating embeddings, and performing similarity search via Weaviate.

Modeled a domain as a formal ontology, serialized it in Turtle format, and served it through Apache Jena Fuseki via Docker for SPARQL-based querying.

Built a containerized FastAPI backend exposing three typed REST endpoints that integrate NLP entity extraction, knowledge graph queries, and RAG retrieval into a single service.