Ahmed – Golang, AI, Kubernetes

Worked on applying LLM for financial news datasets to extract relevant information.

• Implemented a pipeline to annotate and label data using spaCy for semi-automatic labelling;
• Created financial news datasets and used them to fine-tune BERT models;
• Trained and deployed Deep Learning models trained on data;
• Created MLOps pipeline to batch process data using Modal in a serverless paradigm.

Technical interviewing and assessment platform providing skills assessments and AI-powered learning tools to empower high-performing teams to go beyond skill gaps and help individuals cultivate the skills they need to level up.

• Took leadership of the Data overseeing ML and Data related topics;
• Developed and refined comprehensive assessments for evaluating professional skills in machine learning and data science, as a hands-on Senior Machine Learning Engineer;
• Designed industry-level machine learning evaluations with an emphasis on MLOps skills, aligning assessment criteria with evolving industry standards;
• Continually updated assessment protocols to ensure accurate measurement and relevance to current industry practices.

The project involved a leading global provider of space-based data, analytics, and space services and was focused on providing high-resolution weather prediction.

• Delivered a PyTorch-based, HPC-enabled, production-grade weather forecast system;
• Spear-headed a research project with the European Space Agency for innovative regional forecasting methods;
• Executed a high throughput geographical data ingestion project;
• Created a Spark-enabled job to process a high volume of streaming data;
• Set up MLOps on Databricks, using MLfow and experiment tracking to monitor models and new training;
• Created API endpoints for ML models using MLflow.

A GPS high-frequency tracker specifically tailored for the Isle of Man TT bike race.

• Created a pipeline ingesting data from trackers using the LoRaWan protocol through TheThingNetwork;
• Added capabilities for anomaly detection on stream of data.

R&D activity regarding prototyping and testing machine learning functionality for a specific product aimed at boosting the development of financial software.

• Worked on Detecting Drift in Chemical Batch Processes using a Deep Learning approach based on the RNN-LSTM model, incorporating tweaks to enable unsupervised learning;
• Contributed to Multivariate Time Series Reduction, Clustering, and Classification tasks by engineering modified techniques such as 1-NN DTW and Shapelets recognition;
• Developed Sequential Event Patterns Recognition and Prediction capabilities leveraging the WINEPI algorithm;
• Deployed Apache Spark pipelines on Microsoft Azure infrastructure to apply MLlib algorithms across diverse datasets efficiently;
• Revamped the KnowledgeNet software, integrating new data science and machine learning algorithms while enhancing the user experience.