S. Mortezah AI & Machine Learning Portfolio

TOON is a deterministic, human-readable serialization format designed for Large Language Model (LLM) contexts, such as prompts, RAG pipelines, and agent protocols. It targets a 30–60% token reduction compared to JSON for uniform tabular data while remaining compatible with JSON's data model (strings, numbers, booleans, null, objects, and arrays).

Data anonymization is the process of protecting private or sensitive information—such as personally identifiable information (PII) and protected health information (PHI)—by removing or encrypting identifiers. The goal is to preserve privacy and confidentiality while maintaining the data's integrity and usability for analysis, testing, or research.

TOON is a specialized data format designed to be token-efficient when processed by Large Language Models (LLMs), reducing the number of tokens required to represent information.

Learn the fundamentals of hashing and how to implement hashing mechanisms using Python. Detailed guidance is available in the Hashing documentation.

Apache Parquet is a columnar storage format designed for big data processing, commonly used within the Hadoop ecosystem. Unlike row-based formats such as CSV and JSON, Parquet's columnar structure improves performance and efficiency for large-scale data processing tasks.

Hashing is a fundamental data structure technique used to map data of arbitrary size to fixed-size values. Detailed guidance on the concepts of hashing and how to implement them using Python can be found in the Hashing documentation.

Bloom Filters are used for efficient membership testing of sets, allowing a system to quickly check if an element is a member of a set with a possibility of false positives but no false negatives.

Apache Airflow is an open-source platform used to create, schedule, and monitor workflows. In data engineering, it is used to manage complex ETL (Extract, Transform, Load) pipelines by defining workflows as Directed Acyclic Graphs (DAGs). In a DAG, nodes represent individual tasks and edges define the dependencies between those tasks. Airflow supports task retries, email notifications, and various external system integrations.

Keras Core (which evolved into Keras 3.0) is a complete rewrite of the Keras codebase featuring a modular backend architecture. This allows developers to write Keras workflows and run them on multiple deep learning frameworks, specifically TensorFlow, PyTorch, and JAX, eliminating the need to be locked into a single framework silo.

PEP 8 is the official style guide for Python code, providing a set of guidelines and best practices to ensure Python code is readable, consistent, and maintainable across different projects and developers.

Polars is a fast Python library specifically designed for processing and manipulating large datasets, providing high-performance dataframes.