
Name : Ishita Saraf
Profile : Software & ML/AI Developer
Email : ishita201@gmail.com
Phone : +65 85257113
About Me
Hello!👩🏻💻 I'm Ishita Saraf and I am a final year Computer Engineering student at Nanyang Technological University, Singapore. I'm pursuing a specialisation in Artificial Intelligence and Cybersecurity.
As an engineer, my pasion lies in problem-solving, research, and the development of innovative technologies.
I am deeply committed to using my skills in the technological realm to enhance people's lives and make a positive impact.
I aspire to be a part of an organization which not only empowers me to create a difference in the world, but also challenges me everyday to learn and grow.
In the last couple of years, I have gained experience in the fields of ML/deep learning and full stack software development, and I am always on the look out for opportunities that will help me expand my knowledge in these fields.
Working at both start-ups and a renowned software giant like SAP, I have mastered the ability to learn quickly, adapt to thrive in diverse environments, and collaborate within cross-cultural teams.
Skills
My Other Skills: Keras, MATLAB, Firebase, CUDA programming, Code Documentation, Database design, API Development, Scrum development, Machine Learning, Deep Learning, Object Detection, Natural Language Processing (NLP), Verilog, Assembly Language, Embedded C, Shell scripting.
+
SKILLS ACQUIREDWork Experience
My internships at start-up companies as well as a multi-national company have helped prepare me for a demanding career in tech. They have given me the opportunity to hone my technical and inter-personal skils.
Projects
These are some of the projects I'm most proud of and passionate about! To view more of my work, please visit my GitHub
Mobile App Dev and ML
SenSync
A health monitoring and support application for individuals with autism. A mobile app is built using Flutter and Firebase and utilizes wearable watch data to detect sensory overload in autistic users. It uses a 99.2% accurate random forest classifier, and has various features like mood logging, breathing exercises, caretaker connection, real-time alerts during sensory overload episodes, and access to educational material.
Deep Learning
Multi-class Flower Classification
An end-to-end deep learning pipeline which achieved 97.6% mean accuracy in multi-class flower classification using TensorFlow and Keras. It employed transfer learning CNN models and ensemble learning, with hyperparameter tuning for optimal performance. The project also explored few-shot training and implemented a Siamese network with triplet loss for better results with limited training data.
Deep Learning
Decoding Silent Speech from EEG waves
A research project which aimed to build a speech neuroprosthetic which would convert EEG (non-invasive) brainwave data when a patient imagined to speak into actual words. An implementation of a novel study which employs a 4 stage hierarchical model composing a CNN, LSTM, deep autoencoder and XGBoost classifier to classify epoched brainwave data into simple words. Built using PyTorch, this project includes raw EEG data processing, training and testing.
Full Stack Web Dev
STARS++
A MEAN stack web app that is an improved course registration portal for NTU students. It includes automatic timetable generation and course index changing provisions built using graph matching and backtracking algorithms, and an email notification system for important updates.
Data Science
Covid-19 Time Series Forecasting
A graphical CLI application that gives predicted values of crucial figures for the next 15 days amidst the pandemic. Performs data processing, exploratory data analysis and ARIMA time-series forecasting using Python libraries on the Covid Tracking Project dataset.
Cryptography
Attacks on RSA cryptosystem
An implementation of a weak RSA public key cryptosystem in Python to showcase various mathematical security attacks like Hastad Broadcasting Attack, Franklin-Reiter Related Message Attack, Weiner's Attack, and Pollard's factorization attack.
Object Detection & RPi
Object Navigation Robot
A group project in which a robot was programmed to traverse a maze of obstacles. A Raspberry Pi was configured for communication with an Android controller and a server running certain algorithms. Image recognition involved dataset creation, data augmentation, and implementing transfer learning with YOLOv7 model in PyTorch.
Java & OOP
Restaurant Reservation and Sale System
A non-graphical console application for making reservations, recording orders, and displaying sales records at restaurants using Java language and Object-Oriented Design Principles.
Hardware Programming
Path traversal Robot
Code for programming an ARM Cortex M4F-based robot in C to autonomously follow a black-colored path. It includes configuration code and programs for using reflectance sensors, bump sensors, IR sensors, tachometer as well as UART to achieve this.
Leadership
Along with developing my technical skills at university, I was keen on honing my leadership and interpersonal skills. Each of the below roles provided me a unique experience and a unique set of skills!
NTU Open-Source Society - Hackers for Charity Chairperson
Collaborated with charities to design yearlong projects that will help them effectively in their activities.
Lead and supervised building of payment and invoice generation automation software
and website revamping for two prominent NGOs.
Blockchain at NTU Club - Treasurer
Advised club on all financial matters. Spearheaded, maintained and reported all financial transactions of club. Sourced vendors, headed sales and managed inventory for all club merchanise. Prepared and presented budgets for various club events comprising 100+ club members. Secured sponsorships worth S$6000+ for several club activities.
Google Developer Student Club @ NTU Chapter - Core Technology team member
Obtained and organised content for several workshops for club members. Planned and conducted workshop to teach 35 members web application development using ReactJS.
27th Hall 8 Council - Recreational Director
Collaborated with a planning committee of 50+ members to organise Inter Hall Recreational Games (IHRG). Directly managed a subcommittee of 10 members as manpower for running events. Communicated with 20+ companies to attain sponsorships and coordinated multiple events for hall residents.