Shimone Vyas

Skill Nation Certified AI Power BI enthusiast || 2x Salesforce Certified || SFDC QA : Ex-Deloitte || Fostering progression in Software Solutions || Industry exposure on Data Science, Analytics, Computer Vision, Neural Networks ||

SFDC-based QA Engineer possessing 4 prestigious Awards previously with Deloitte USI as a Team Lead.

Certified Salesforce Administrator, AI Associate. Consulting, Salesforce CRM. Data Science industry experience.

Certified AI-based Power BI based emerging Data Scientist.

• Skill Nation : AI based Microsoft Power BI Dashboards (07/2025)
• Salesforce : Salesforce AI Associate (04/2025)
• Salesforce : Salesforce Administrator (06/2023)
• Tricentis TOSCA : TOSCA Automation Specialist 1 Tricentis (10/2021)
• Tricentis TOSCA : TOSCA Automation Specialist 2 Tricentis (10/2021)
• Udemy : Testing in Agile (09/2021)

Project: Analysis of Neural Network Architecture and Machine Learning for Emotion Detection Duration: May 2024 – Sept 2024

💠 Engineered Advanced Models: Reliable models for emotion detection were constructed employing
KNN, CNN, and Random Forest algorithms against different emotions.

💠 Optimized Performance with Hyperparameter Tuning: The model's efficacy was enhanced by
complying with industry best practices and achieving superior performance metrics. Strategic
hyperparameter tuning was leveraged by altering parameters namely Epochs, number of layers,
kernel size, number of filters, k-value, nature of optimizers, and activation functions.

💠 Integrated Cutting-Edge Research: Performed thorough analysis of cutting-edge Computer Vision
research, and literary works, incorporating current trends to improve project outcomes.

💠 Employed Advanced Optimizers and Activation Functions: To increase learning effectiveness and
model correctness, CNN models were fitted with optimizers like ‘Adam’, ‘RMSprop’, and ‘SGD’ in
addition to activation functions like ‘ReLU’ and ‘Sigmoid’.

💠 Applied Powerful Libraries: Streamlined development processes and improved analytical capabilities
by working effectively with key libraries namely NumPy, Keras, TensorFlow, RandomUnderSampler,
Scikit-Learn, etc.,

💠 Evaluated Model Performance with Precision Metrics: The framework’s proficiency was evaluated
utilizing crucial metrics like Accuracy, F1-score, Re-call, and Precision to guarantee thorough
performance evaluation. Confusion Matrices and graph plots yielded visually meticulous findings.

💠 Comprehensive Comparative Analysis: Performed comprehensive comparative review on bifurcated
emotion and multi-class datasets, providing insights that guided model selection and improvement
tactics. Multi-class yielded 26% upon applying the ‘Tanh Activation’ whilst Binary bifurcated class
yielded 64% for ‘Tanh’ and ‘ReLu’ Activation functions and 63% for the ‘Sigmoid’ Activation function.

Deloitte USI :

• Applause Award, Nov 2022
• Applause Award, Jun 2022
• Spot Award, Jun 2021
• Spot Award, Jan 2020