Harsh Shah

Skills

Data Analytics

Generated Analytical Insights from the data by extracting meaningful information from data to support decision-making using Python libraries and Data Visualisation Tools (PoweBI).

Machine Learning

Hands-on experience in developing Machine Learning and Deep Learning models on diverse datasets and achieving State-Of-the-Art accuracy.

Natural Language Processing

Experience in Training and fine-tuning Language Models and developing modern transformer-based architectures for various uses-cases of NLP.

Software Develoment

Worked on Software Development / Android projects for the back-end developments, bug fixing, maintenance and improvement of applications performance

Database Management

Monitoring secure connection to a relational database, managing databases, retrieving data according to the query submitted by the user.

Computer Vision

Comfortable with Digital Image Processing, Visualisation of Images, Image Classification and Object Detection using Deep Learning with Python

Projects

Hardware Store Sales Analysis

Sales analysis for the Hardware Store

Tracking of the revenue through the timeline and zones analysis to see the trend and pattern for the change in revenue.

Showed the comparison between two customer markets to show the contribution by different customer types

Created dashboard reports using PowerBI, Setup workflow with MySQL workbench and queried in the database to provide insights to support the decision-making

Data Analysis ETL SQL PowerBI Data Modelling

GitHub Link →

Argument Snippet Generation

Optmisation in textual snippet generation for an argument search engine

Used the graph-based extractive text summarization method inspired by Google’s PageRank algorithm to improve the accuracy by 9% over than previous snippet generation approach.

Trained the language model on Argument dataset using transfer learning technique to enhance the readability and representativeness of snippets in the search engine

NLP Deep Learning Machine Learning Search Engine Optimisation Transfer Learning

Github Link →

Data Extraction, Search, Analysis and Benchmarking

Extracted training data from knowledge graph files and parsed it on the data extractors to automatically identify the best extractors response for the given set of sentences.

Transformed the unstructured extractor's responses in the unified RDF triple formats to analyse the responses to extract knowledge.

Developed a Machine Learning model (AutoML, Random Forest Classifier, Decision trees) to predict the best extractor for the given category of text data.

Data Exploration NLP Data Analysis Machine Learning Project Management

GitHub Link →

More projects

Research Papers

Applications of Search-based Software Engineering (SBSE)

Research paper highlights the most widely applied search techniques in SBSE in differnt stages of the software engineering life cycle from requirements, design, testing, project planning to project maintenance and re-engineering. This paper also provides an analysis of different application areas of SBSE in Testing and search techniques applied in these areas.

View Research Paper →

Optmisation in snippet generation for an argument search engine

Motivated by the need for specialised snippets in an argument search, this research paper explores different approaches to enhance the groundwork on argument snippet generation. Modelling Context of an argument by finding the other arguments addressing similar controversial issues and modifies the argumentative computation method with the use of transfer learning techniques to bias the selection of snippets to be more argumentative.

View Research Paper →

Hello 👋

I'm Harsh Shah