Data Scientist
2017 - PRESENTSatalia- Built a convolutional neural network that predicts spatio-temporal grocery deliveries demand. This improved the performance of a last-mile delivery optimization algorithm and enabled businesses to dynamically price deliveries based on the forecasted cost.
- Conducted various consultancy workshops to assess technological needs and facilitate the adoption of artificial intelligence solutions in industries such as finance and retail.
- Researched and developed a proof-of-concept algorithm to determine the best solver algorithm for a given optimization problem to be embedded as a machine learning layer of on-demand optimization software. My research focused on computer science problems known as Automatic Algorithm Selection and Configuration.
- Produced three machine learning models for a large UK retailer: Recurrent neural network to predict daily demand, dense network to predict the hourly distribution of daily demand, and an optimization model to recommend hourly workforce allocation. This enabled the retailer to forecast hourly demand with 80% accuracy.
Technologies: Python, Keras, TensorFlow, Google Cloud Platform, OptimizationFounding CTO
2016 - 2019PigeonLine- Constructed the company’s technical foundations, which involved architecting and deploying machine learning algorithms, back-end APIs, front-end interfaces, and the server infrastructure. To achieve this, I jointly managed a team of four developers.
- Developed an interactive topic analysis web application using neural word embeddings.
- Designed and delivered an introductory course on AI and Social Sciences for the London School of Economics. It used our interactive topic analysis tool to teach students how Natural Language Processing can aid Social Science research.
- Produced in collaboration with the New York University of Abu Dhabi, a tool to fetch, parse and organize over 700GB of data (including PDF attachments) from the US regulations API to study citizen sentiment and predict public unrest.
- Raised $1 million, alongside the CFO and CEO to finance business expansion, technical development, and product growth.
- Researched, developed, and deployed an NLP matching algorithm that automatically sourced the most suitable civil servants for new government projects for a Prime Minister’s office in the Gulf Region.
- Won the 2018 LSE Entrepreneur-of-the-Year Award and Distinguished Alumni of Dubai Future Accelerator.
Technologies: Python, Django, JavaScript, Vue.js, PostgreSQL, SQL, Docker, Spacy