Close

Akshay Bajpai


M.Tech (Communications & Signal Processing)

Electrical Engineering, IIT Bombay

Resume

About Me


Post graduate in Electrical Engineering

Wireless Communication and Deep Learning Enthusiast

System Administrator, Electrical Engineering Department, IIT Bombay


Experience

IIT Bombay, Mumbai

Institute Research Assistant (IRA), Electrical Engineering Dept.

  • Pursuing M.Tech in Communications & Signal processing as IRA assigned to PCLAB, Electrical Department
  • System administrator with primary focus on developing web based applications for digitization and smooth functioning of the department
  • Masters thesis on Optimal precoder estimation and interpolation for MIMO-OFDM systems.
  • Also working on multiple deep learning based histology application projects in MeDAL(Medical Deep Learning and Artificial Intelligence Lab) at IIT Bombay

Wipro Technologies, Mumbai

SOC Analyst

  • Performed real-time monitoring,investigation and mitigation of security incidents for multiple clients, both Indian and International, with network spanning all over the country.
  • Created new rules based on security use cases on HP Arcsight and created informative dashboards and reports enhancing the teams’ capability to find and prevent security breaches pro-actively.

Directorate of EOTS, DRDO, Balasore

Intern

Title : Numeral Extraction and Recognition
Guide : Dr. Pradipta Roy,Scientist E, ITR, DRDO, Chandipur

  • Implemented algorithm to extract elevation and azimuth angle values from videos of real missile testing captured by Electro Optical Tracking Systems.

School of Mechatronics & Robotics, IIEST Shibpur, Kolkata

Research Intern

Title : Pressure Mapping and Alert System for Posture Assessment for Paraplegic Patients
Guide : Prof. Subhasis Bhowmik

  • Developed device to aid paraplegic patients in order to prevent pressure sores while using wheelchairs.
  • Involved usage of GSM and bluetooth for developing mobile application to aid paraplegic patients. Poster available on research gate.

Education

Indian Institute of Technology, Bombay (IIT Bombay)

July 2019 - July 2022

Masters of Technology in Communication Engineering

CPI - 9.59

Autumn (2020-21):
I Stage Project: Optimal precoder estimation and interpolation for MIMO-OFDM systems, R & D Project

Spring (2020-21):
Wireless Communication, Information Theory & Coding

Autumn (2020-21):
Digital Image Processing, Speech Processing, Advanced Topics in Machine Learning, R & D Project, Seminar

Spring (2019-20):
Communication Networks, Introduction to Machine Learning, Network Security

Autumn (2019-20):
Statistical Signal Analysis, Digital Signal Processing and its Applications, Digital Message Transmission

National Institute of Technology (NIT), Durgapur

August 2014 - June 2018

Bachelors of Technology in Electronics & Communication Engineering

CPI - 8.97

Kendriya Vidyalaya Cossipore, Kolkata

2014

Higher Secondary - CBSE

Percentage: 97.00%

Kendriya Vidyalaya Cossipore, Kolkata

2012

Secondary - CBSE

Percentage: 95.00%

Projects - Communications

Optimal precoder estimation and interpolation for MIMO-OFDM systems

  • Reviewed and compared existing techniques to represent precoder matrices using Unitary, Grassmanian, Stiefel and Flag manifolds for efficient estimation and interpolation
  • Simulated temporal and frequency interpolation on flag manifolds in C++
  • Implemented gradient descent algorithm to find geodesic between two flag manifold elements to maximize rate with minimal feedback

View Report

Multi-User Massive MIMO: Benefits and Challenges

  • Studied the effects of antenna count on achievable communication rate and the practical challenges associated
  • Performed simulations for MISO, SIMO , MRC , VBLAST receiver & Alamouti Coding to obtain plots for achievable rates using ITPP library in C++
  • Implemented single-carrier communication to overcome high PAPR requirement in case of MIMO-OFDM

View Report

Degraded Broadcast Channels and Superposition Coding

  • Reviewed capacity bounds of discrete memoryless channels with focus on degraded broadcast channels (DBC)
  • Studied codebook generation techniques using superposition coding to attain achievable rate for DBC

Implementation of Routing Information Protocol (RIP) using C

  • Implemented RIP protocol in C usingsocket programming (Bellman Ford Algorithm)on a Linux PC
  • Capability to handle addition and removal of nodes from the network dynamically

Projects - Deep Learning & Signal Processing

Robust Deep Learning Framework to address General Label Noise in Medical Imaging

  • Designed a supervised contrastive learning framework by forming training curriculums with the help of class prototypes on colorectal cancer dataset
  • Developing a novel iterative re-weighting scheme for robust deep learning classification under open set label noise for medical imaging data

View Report

Python-SQL based UI for Zero-Based Timetabling project for Indian Railways (IR)

  • Developed UI for efficient analysis and correction of Zero Based Time Table (ZBTT) simulator output
  • Enhanced error detection and correction speed by many folds by creating UI for analysing the ZBTT simulator outputs using plotly and SQL. The tool was given to Indian Railways as an add on to original deliverable
  • Streamlined the project using version control and makefile

Image Segmentation of Right Heart Ventricle

  • Engineered Pixel-wise Segmentation of Right Ventricle of Heart in MRI Images from RVSC-MICCAI 2012 dataset with UNet Architecture as baseline model to get a Dice-score of 0.2761
  • Performed ablation studies with hyperparameter tuning to achieve 0.24 Dice-score with Switching loss

PyQT based UI to train ML models

  • Developed UI using PyQT to enable quick data visualisation and ML model training
  • Implemented several models (Linear regression,SVM,Logisitc regression, Neural networks,Decision tress, Random forests) that could be tuned, trained and saved on the input data

LPC Analysis of Synthesized vowels & Natural Speech

  • Synthesized phonemes by implementing source filter model to replicate the human glottal source
  • Performed Linear Predictive Analysis of natural and synthetic speech which involved LP coefficient estimation, constructing LP magnitude spectrum and sound reconstruction

Digital Photography with Flash and No-Flash Image Pairs

  • Implemented de-noising and detail transfer to merge the ambient qualities of the no-flash image with the high-frequency flash detail, using cross-bilateral filtering
  • Performed white-balancing to change the color tone of ambient images, continuous flash to interactively adjust flash intensity, and red-eye removal to repair artifacts in the flash image

Evolution of Human Pose Estimation

  • Reviewed the state of the art method of human pose estimation which propagates known person locations forward and backward in time and searches for poses in those regions.(combined detection and tracking)
  • Studied multi-timescale trajectory prediction for long term abnormal human activity detection

Sentiment Analysis on IMDB dataset

  • Trained LSTM, GRU, Bi-LSTM and Bi-GRU to study performance changes with number of units and layers and acheived an accuracy of 87% BiGRU
  • Studied the effects of using different learning-rate schedulers like Cosine Annealing scheduler, Exponential rate scheduler, Step scheduler and Reduce Learning Rate(LR) on Plateau scheduler

Employee Attrition Classification

  • Classified on Employee Attrition Kaggle dataset by conducting data analysis and various mix of classifiers
  • Obtained the best test accuracy of 90.09% on the test set with GradientBooster classifier

Support Vector Machine Classifier from scratch

  • SVMs are a set of supervised learning methods used for classification, regression and outliers detection
  • Implemented and trained SVM Classifier using gradient descent method and visualised it in Python
  • Visualised the output decision boundary of the model trained on linearly separable and non-separable data

Skills

-->