Machine Learning Engineer

Machine Learning Engineer || Co-founder at 404Enigma || Researcher @SCMIA || Blogger at Medium || Kaggle Contributor || Full Stack Developer Django

profile_img

DeskApp Software

Desktop 🖥 application based on AI. A productivity tool, which uses AI to analyze user activity.
So the fundamental idea behind the project is to use our own data to study our own behavior and optimize our work efficiency by leveraging the power of AI algorithms and analytics.
Users can learn where are they spending their time most and also compare the amount of work done per day for months.
It has different graphs to show our performance.

See Work

profile_img

Cityscape Semantic Segmentation

Performed semantic segmentation on Cityscape Dataset.
Used variational autoencoders(VAE) and UNet Architecture to construct or segment the input image.
UNet gave better results as compared to VAE.

See Work

profile_img

Machine Learns

An online platform for learning and visualizing machine learning and deep learning concepts.
In this, the trained model is hosted on the cloud, users can interact with the model and learn how it works in real-time.
Presently It demonstrates how VAEs perform image compression, and how noise from images is removed using VAEs.

See Work

profile_img

SELF DRIVING CAR

This is my first Reinforcement Learning project.
I have built a self-driving car simulation where we can draw paths and the car learns to pass through them.
I have used deep Q-Learning. Have learned concepts like Exploration & Exploitation, Experience Memory, Reward setting, etc.
Kivy is used to show the simulations of the car.

See Work

profile_img

Fall Back To Safe Zone

This project demonstrates how to smartly manage a large-scale disaster (ex 19|11 attacks ).
This is a concept that considers factors such as injured individuals, rescue teams, and people who are unaware of the threat and locates the safest zones, such as hospitals, and distributes the injured into them.

See Work

profile_img

Sorting Algorithm Visualizer

This is a web application that Visualizes how the sorting Algorithm works along with its Time Complexity.
It also shows the different steps involved in the sorting algorithm.

See Work

profile_img

SEARCH WORD BOT

It takes Screenshots of the given Search Word and their Answers.
Applies OCR to extract all letters and Answers.
Have used tesseract and cv2 library to perform OCR and preprocessing.
Then applies the Algorithm to search answers and displays them using Tkinter.

See Work

profile_img

2048 BOT

Play 2048 On the Computer and Beat The BOT if YOU Can.
This is a bot that tries to play 2048.
It basically uses an algorithm to estimate the possible rewards by making a move in different directions and based upon that it selects the best action to take.

See Work

profile_img

Brain Tumor Segmentation

In this project I created a semantic segmentation model using the PyTorch framework called MONAI.
I have applied various data augmentation techniques and have built a UNet deep learning model.
Objective was to learn to use the MONAI framework for applying deep learning to medical images.

See Work

Brain MRI Synthesis (Image Translation)

In this project T1- weighted brain images are converted to T1-weighted images.
I have used variational autoencoder (VAE) architecture to map the conversation from T1 to the T2 modality.
Have achieved a maximum of 0.15 RMSE on the validation dataset.

See Work

profile_img

Real Time Digit Recognition

This is a fun project in which I have created a deep learning model to recognize handwritten digits.
In this users can write digits on a Tkinter-based interface which is then preprocessed and passed to the dl model.
The prediction along with its accuracy score is displayed on the Tkinter interface.

See Work

profile_img

Customer Segmentation

In large business firms, the major aspect to deal with is a large number of feedback from the customers.
In this I have used the Snapchat feedback dataset.
The objective was to categorize similar types of feedback given by the Snapchat users.
I have to build a clustering model to capture similar types of feedback.
I have also applied Topic Modeling.

See Work

profile_img

ModelAuto 📚

Pip install ModelAuto ( ML Automation Library ).
It is a Python library used for automating machine learning model-building tasks.
It provides us with helpful tools to perform tasks like Data preprocessing, Feature selection, and Model Selection with few lines of code.

See Work

I enjoy creating products and applying algorithms to various fields, particularly those that employ ML or AI concepts. On Medium, I publish machine learning blogs. I enjoy participating in hackathons because we get to work on real-world challenges, and have won a number of them as a team. I am currently an intern as a Machine Learning Engineer at Intellekt AI, where I am having experience with reinforcement learning in the finance domain. As a Researcher at SCMIA, I gained experience with medical images and computer vision. In addition, I worked as a data analyst at Appopener. I am a co-founder of 404Enigma where we develop products.

Skills

PyTorch
Tensorflow
MONAI
Computer Vision
NLP
Reinforcement Learning
Django
Java Script
Flask
AWS Elastic beanstalk
Firebase
Blogger
Kaggle Contributor

Machine Learning Engineer

I worked on using Reinforcement Learning (Artificial intelligence) in the automating trading. Here a reinforcement learning agent predicts when and how much to buy sell or hold stocks using PPO algorithm.
I learn working on large code bases and properly utilizing the concepts of Object-oriented programming to organize the code such that it becomes reusable and more Independent.
I also learnt importence of creating Testing. I created test cases (Unit testing) using PyTest in python.

Research Intern

A research project which involves using deep learning algorithms to help in the medical image to image translation.
The research involves conversions of MRI images to CT images using deep learning algorithms.
Here I have successful worked on this project in converting the MRI to CT images using Pix2Pix, Unet, VAE.

DATA ANALYST

It is a web-based application for content creators and influencer for creating deep links.
My work here was to handle the data analytics on the website, Creating an Analytics dashboard, using Django and Analyzing the growth factors via Google Analytics, Google Tag Manager.

1st 🥇 position in NanaGiants

In large business firms, the major aspect to deal with is a large amount of data. One such aspect is " feedback from the customer". It uses Unsupervised Learning for Customer Segmentation and also performs Topic Modeling

3rd 🥉 position in Cicada

This is a concept that considers factors such as injured individuals, rescue teams, and people who are unaware of the threat and locates the safest safe zones, such as hospitals, and distributes the injured into them.

1st 🥇 position in Sociothon

This project tackles the problem of a weak supply chain. We have used Deep Learning to predict the sales of drugs for a particular month. Our platform is a positive means to spread awareness of the ongoing Govt. scheme JAS.

First Seminar Conduted

As the Co-Lead of Machine Learning & Artificial Intelligence, I conducted my first seminar on "What is Machine Learning About". In this seminar, I have covered how all the ML & deep learning algorithm is about finding the best fit line and more.

Email Me:

sudhanshu.dpandey@gmail.com

Follow Me: