Siddharth Sanghavi

Mumbai, India

Throughout my experience as a Computer Engineering undergrad, I've developed interests in the areas at the nexus of Computer Vision, Robotics, IoT and Deep Learning.

I enjoy applying my technical skills to develop applications that help people automate boring tasks and assist computers in seeing, moving and understanding like humans.

When I'm not preoccupied with Computer Science stuff, I like to indulge in Calligraphy, Sketching and Painting.

Let's collaborate on a project. Reach out to me via:

Education

Dwarkadas J. Sanghvi College of Engineering

Bachelor of Engineering, Computer Engineering

CGPA: 9.97/10 (in coursework till now)

Senior Mentor & HR head at DJ Unicode

Aug. 2017 - June 2021


Pace Junior Science College

Maharashtra State Board of Secondary and Higher Secondary Education

Finals: 89.91%

JEE Mains Rank: 31000

July 2015 - June 2017

Experience

Robert Bosch Centre for Cyber-Physical Systems @IISc Bangalore

Research Intern

  • Currently, I am working on the problem of Dynamic to Static Translation (DST) of LiDAR scans using Deep Learning for graph-based SLAM.

Jan. 2021 - Present

Lido Learning

Full Stack Web Development Intern

  • Designed and developed a passwordless user authentication microservice for their E-Learning Web Platform using Node.js, AWS Lambda functions and ElastiCache for Redis.
  • Developed a feature for analytics involving tracking of user activity on their E-Learning Web Platform using React and Hasura GraphQL.
  • Collaborated with the Backend team to deploy an Apollo GraphQL server.

July 2019 - Sept. 2019

Projects

A collection of my personal and collaboration projects that I've undertaken over the course of my undergraduate education to explore different fields in Computer Science and narrow down my areas of interest.

Eeg2Vec (currently ongoing)

Since every person possesses a unique combination of biological brain structure and involuntary memory, EEG-based Biometric Authentication is more difficult to duplicate or steal than fingerprint and facial biometrics. This project aims to utilize EEG-based Biometrics to authenticate a person’s identity using Brain-Computer Interfaces (BCIs) and Deep Learning.

Code TensorFlow Keras MNE-Python NumPy Matplotlib

SpatioTemporalNet

This project aims to classify the sleep stages which will help to gain a better understanding and diagnosis of sleep disorders such as sleep apnea, insomnia, etc. Therefore, we present a comparative study of two Deep Learning architectures used to classify the sleep stages based on EEG signals.

Code TensorFlow Keras MNE-Python NumPy Matplotlib

MonoDepthNet

Single View Depth Estimation is pivotal for inferring scene geometry from an RGB image. With the advent of deep neural networks, Monocular Depth Estimation has achieved promising performance in accuracy. This project aims to leverage Transfer Learning and structural dissimilarity measures to produce less blurry depth maps.

Code TensorFlow Keras OpenCV NumPy Matplotlib

e-Yantra Robot

My team and I build an autonomous path following robot under the "Thirsty Crow" theme for the IIT Bombay e-Yantra Robotics Competition. The robot had to find the shortest path to objects placed on an arena and optimally pick up and place them on designated spots using its electromagnetic arm and an overhead camera feed.

Code Python Embedded C OpenCV ATmega2560

Neural Style Transfer

Neural style transfer (Gatys et al.) is an optimization technique used to generate an image with the same "content" as a base image, but with the "style" of a different reference image (such as a painting by a famous artist). This project aims to create art using Convolutional Neural Networks.

Code TensorFlow Keras OpenCV NumPy Matplotlib

Image-based Search

Content-based Image Retrieval (CBIR) is used to searching a database for images that match a query image. This project uses a Convolutional Autoencoder to form latent image embeddings. A data structure called vantage point tree is used for storing and retrieving those embeddings in O(n * log n) and O(log n) time respectively.

Code TensorFlow Keras OpenCV NumPy Matplotlib

Master's Information Portal

I lead and guided a team of 5 of my juniors to develop a Yocket-like website for connecting the students of my department with alumni who have pursued MS in CS abroad. This website has features like a forum, chatrooms and tags to help the students network with alumni and ask them queries regarding the graduate admission process, universities and courses.

Code Node.js React

Attendance Manager

As a junior, my team and I built an internal website to automate the student attendance management system in my department. It has features to track student attendance and automatically generate excel sheets and reports for the faculty members.

Code Django REST React

Event Manager

As a junior, my team and I built an internal website to automate the documentation of event reports in my department. It has features like a dashboard calendar to reserve slots for events and activites and automatically generate PDFs based on the description of the events by the faculty members.

Code Django REST React

Publications

Smart Shopping Cart

We propose an IoT-based human following shopping cart to provide people with a hassle-free and interactive shopping experience. Our approach allows the cart to follow its user without line-of-sight sensors and line following techniques. In addition to that, we have integrated a web user interface and RFID Reader to keep a track of the products in the cart for budgeting and billing purposes.

Paper Code Node.js React MQTT Arduino NodeMCU

Skills

Technical Skills

Here are a few technologies I've worked with or implemented in my projects, coursework and hackathons. Hope to add some more technical skills to this list in the future!

Programming Languages

Python JavaScript C++ Golang

Libraries & Frameworks

Keras TensorFlow OpenCV NumPy Matplotlib Seaborn Arduino

Database Technologies

MySQL PostgreSQL SQLite MongoDB Redis

Web Development

Node.js Serverless GraphQL Django React.js Bootstrap4 HTML5 CSS3

DevOps

Git Docker AWS Google Cloud Heroku Hasura

Operating Systems

Linux Windows



Leadership & Collaboration Skills

Senior Mentor, HR Head & Backend Developer @DJ Unicode

  • DJ Unicode is a student driven technical committee in the CS department which promotes open-source development to students. With an aim of knowledge transfer and community building, DJ Unicode has successfully grown from 45 developers in its first year to 80 developers in its second term with 20 ongoing projects.
  • I worked as Student Developer at DJ Unicode from October 2018 to August 2019, my team and I developed two internal websites for automating the Attendance and Event Management for my department. These websites automated the task of manually documenting student attendance and event reports.
  • Currently, I'm the Senior Mentor and HR head at DJ Unicode since August 2019. I lead and advise ~50 juniors for incorporating Backend Web technologies like Node.js, Django and MongoDB in other in-house projects. Also, I mentored a team of 5 juniors for developing a Masters Information App to help the students of my college connect and network with Alumni.

Member of Sponsorship & Organising Committee @ICACTA 2020

  • My team and I identified potential sources of funds for fundraising and sponsorship and handled clients including donors, grant agencies and sponsors.
  • I worked as a volunteer to oversee smooth conduct of the research papers presentation sessions.

Blog

My attempt at putting the various experiences I've had till now into words. More to come soon!

Internship at an EdTech Start-Up

Recounting my days as a web development intern at Lido Learning right from my interview to the lessons I learnt while working on their online platform.

Link