Deep Research Data Scientist
WorldQuant is a quantitative asset management firm with more than 800 employees spread across 28 offices in 17 countries focused on
developing high-quality financial strategies across a variety of asset classes in global markets, utilizing a proprietary research
platform and risk management process.
Time series analysis, nowcasting, alpha development.
Building tools for evaluating alpha performance and risk.
Processing, transforming, analysing large alternative datasets using Spark.
Full stack machine learning model development (Keras/TensorFlow, Flask, Celery, Redis, Docker, Kubernetes)
Machine Learning Engineer
Aiden is virtual NLP-powered assistant that monitors mobile app campaigns, generates and implements approved recommendations, and tracks their impact.
Built multi-task deep learning-based ad spend forecasting models to answer questions such as “What will be the total number of impressions, link clicks, and app installs next week if I spend $X on ads ABC targeting Y audience in Z location?”.
Using Flask, Python/R, and Docker, developed and deployed a REST API app to detect anomalies in advertising time series data.
Developed a time-series anomaly detection algorithm.
Backend development, scheduled data download and preprocessing, and model deployment (Python, R, Docker, Flask-based REST API microservices, Celery, Redis, PostgreSQL).
Data Scientist
Picasso Labs is recognised as one of Unilever Foundry’s most ambitious and innovative start-ups of the past 5 years.
It helps brands like Unilever, Louis Vuitton, Four Seasons, Tesco, AB InBev and Samsung understand the visual preferences of their audience,
reduce guesswork, stand out in their industry and improve marketing performance.
Using Keras and OpenCV, developed a convolutional neural network (CNN)-based model with additional input branches for HSV color histograms and entity embeddings of date, time, and image tags for recommending images to improve CTR or engagement on social media.
Re-used the model above to create a content-based image retrieval (CBIR) engine using transfer learning, HOG, and HSV features as indices.
Applied non-parametric statistical tests to determine best performing image categories.
Using Keras and OpenCV, built a custom facial expressions recognition model and applied it on over 5000 web-scraped images from US online media to investigate “visual bias”.
(please see 1 and 2).
University College London (UCL)
Teaching Assistant
Supervised research projects and taught MSc and MEng students Discrete-Event Simulation, Mathematical Programming (LP and MILP), Genetic Algorithms, and Multi-Objective Optimisation.
Sphere Fluidics Limited
Consultant
Created fluid-flow models of microfluidic chips for a novel single-cell screening and analysis system.
UCL Advanced Centre for Biochemical Engineering
Research Associate
Performed multi-variate data analysis on mass spectrometry data to improve the expression of virus-like particles from Pichia pastoris cells for a universal influenza vaccine project.
