Looking for a job

Menna Osman

@mennaosman

Machine Learning Software Engineer at Pulse by Solutions building production RAG assistants, real-time voice agents, and federated unlearning systems.

Egypt

Message

At Pulse by Solutions (Giza Systems), I architected a real-time voice agent using self-hosted LiveKit and integrated speech-to-text, text-to-speech, turn detection, and interruption handling with a production RAG assistant.

I also engineered the backend for a RAG-based learning assistant and built a modular federated learning system with Flower and PyTorch. Its server-side class-level unlearning approach reduced forget-class accuracy to nearly 1% while retaining accuracy on the remaining classes.

At MBZUAI, I developed computer vision models for medical imaging and trained head-and-neck cancer segmentation models, achieving an average Dice score of 0.763. My work also includes published research and projects in blockchain, emotional voice conversion, and gait analysis.

Experience

Work history, roles, and key accomplishments

Current

Software Engineer – Machine Learning

Current

Pulse by Solutions (Giza Systems)

Mar 2025 - Present (1 year 7 months)

Architected a real-time voice agent using self-hosted LiveKit and built an end-to-end speech pipeline. Engineered backend for a RAG-based learning assistant and developed a federated machine unlearning system.

RAG LiveKit WebRTC Stts TTS Python PyTorch Flower Gan Lang Chain Pinecone Docker PostgreSQL Airflow ELK Stack Power BI Redis NoSQL LTI OpenEdx

Data Scientist Intern

Pulse by Solutions (Giza Systems)

Sep 2024 - Feb 2025 (5 months)

Completed a 6-month intensive training program covering Python OOP, Django, machine learning, and deep learning. Deployed machine learning models across various tasks and built a GenAI-powered recruitment chatbot.

Python Django Machine Learning Deep Learning NLP Time Series AutoML Docker Power BI Airflow ELK Stack NiFi SQL NoSQL Lang Chain Pinecone Streamlit

Machine Learning Intern

Modus.vc

Aug 2023 - Sep 2023 (1 month)

Created an LLM chatbot using Falcon-7B with Lit-GPT for fine-tuning, quantization, and inference. Prepared training data and configured the development environment on a virtual machine.

LLM Falcon 7B Lit GPT Fine Tuning Quantization Inference Virtual Machines

AI Research Intern

MBZUAI (Mohamed bin Zayed University of Artificial Intellige

May 2023 - Jun 2023 (1 month)

Built foundational machine learning models and neural network architectures from scratch. Developed CNNs and Vision Transformers, and processed multimodal PET/CT medical imaging data for head-and-neck cancer segmentation.

Machine Learning Neural Networks CNN Vision Transformer PyTorch Medical Imaging NnUnet Segmentation

Education

Degrees, certifications, and relevant coursework

Egypt-Japan University of Science and Technology (E-JUST)

Bachelor of Science, Computer Science and Engineering

2019 - 2024

Grade: 3.65/4.00

Activities and societies: Head of E-JUST Operation Team; Head of Hult Prize at E-JUST; Semifinalist at Shell Imagine the Future; Member of E-JUST ROV and IEEE E-JUST.

B.Sc. in Computer Science and Engineering with a CGPA of 3.65/4.00. Led the E-JUST Operation Team and Hult Prize at E-JUST, and participated in Shell Imagine the Future and IEEE E-JUST activities.