Edgar Yépez

Data Science and Optimisation

• Mathematical formulation for optimising the sequencing of garment supply to an automated sorting belt in a warehouse. Improvement measured using the Gini index as a workload cost function.
• Development of an application based on random search heuristics using Python to minimise the workload cost function and return optimal sequences of garments.
• Outline for further optimisation of the workload cost function using quantum computing methods.
• Designed and developed a 2D animation tool to simulate the optimal sequencing of garments and track the workload cost function throughout time.

Software Development

• Collaborated with an interdisciplinary team of attorneys, accountants, and systems engineers to develop tools that automated processes in corporate law, intellectual property law, and accounting workflows.
• Development of a tool based on TF-IDF and BM25, using C# and Angular, to construct a phonetic index of registered trademark names. It achieved efficient querying and identification of similarities between existing trademarks and new trademark submission requests, by replacing an O(n²) all-to-all comparison approach with a close to O(1) querying approach.
• Automated the process of tracking payments to company suppliers by designing, developing, and deploying a full-stack web application using C# and TypeScript. Responsibilities included requirement elicitation from stakeholders, writing documentation and user manuals for both technical administration and end users, as well as coordinating and reporting to stakeholders at multiple hierarchy levels across the company and its clients.
• Achieved successful migration of database catalogues for accounting records and corporate data from ERP SAP Business One to the GeniusLegal legal and accounting system. Led requirement elicitation from stakeholders and coordinated with accounting and legal teams. Developed a full backend tool, using C# and Python, to execute migrations and ensure data integrity at storage, schema and reporting levels.
• Integration of the HighQ law platform and the Kira AI engine with ContractExpress to automate the analysis and amendment of contracts.
• Training an intent-based chatbot using IBM Watson Assistant to guide users in establishing new local companies, including the development of an evaluation tool based on the F-measure to monitor the accuracy of the chatbot.

System Administration

• Automating system tasks through scheduling and scripting.
• Administration of Windows Server operating systems, networks, and SQLServer databases.
• Migration from Google Workspace to Microsoft 365, including the development of an application to compare account contents and directories across cloud services.
• Administration of ERP SAP Business One, GrantThornton DBWare accounting system, ThomsonReuters HighQ document repository, Microsoft 365 tenant, and Google Workspace.
• Supplier management and technology inventory control.

Research Intern: Information Retrieval

• stuff i did.

Research Intern: Natural Language Processing

• Development and documentation of a command-line tool to parse and organise corpora for bidirectional English–French translation, composed of over 60,000 HTML documents. Over 500 regular expressions were written and debugged in Perl to parse the morphological structure of the corpora. The resulting structures were then processed using manually designed finite-state machines.
• Detection of French liaison in speech recordings through acoustic feature extraction and phoneme alignment, using language processing libraries such as NLTK and standalone tools like openSMILE, SPPAS, and Praat.
• Classification of speech acoustic signals using support vector machines.
• Development of a Perl library to handle grapheme-to-phoneme conversion pairs.
• Development of a Perl library to handle multi-tier annotation files in the TextGrid format.

Research Intern: Deep Learning

• Training of artificial neural networks for regression using Python to estimate the origin in polar coordinates of acoustic signals emitted in a controlled environment.
• Visualisation of the polar coordinates within spheres representing the origin of the emitted signals.

Teaching Intern

• Development of a command-line application using Python to validate manual data labelling for executing machine learning workflows.
• Development of a command-line application using Perl to check and grade student's programming assignments.

Software Development

• Design and development of a double-entry bookkeeping system to track inventory and the provision of groceries to low-income households. The system was implemented as a standalone desktop application built with VB.NET and MSSQL Server Express.

Help-desk Assistant

• Installation of Windows and Linux operating systems on workstations and servers.
• Configuration of operating systems to enforce internal security policies.
• Installation of office suites, custom internal applications, and anti-malware utilities.
• Preparation of technical documentation and user manuals for internal applications and processes of the Technology Department.