ADITYA SRIVASTAVA
Fifth Year Student E-mail ID: adityasv@iitk.ac.in
Department of Electrical Engineering Mobile Number: 8707400900 /
IIT Kanpur 9450691510
EDUCATIONAL QUALIFICATION
YEAR DEGREE INSTITUTION CPI/%
B.Tech - M.Tech Dual Degree 10.0 (PG)
2014-19(expected) IIT Kanpur
(Electrical Engineering) 9.7 (UG)
2014 CISCE(ISC Board) City Montessori School, Lucknow. 97.5
2012 CISCE(ICSE Board) City Montessori School, Lucknow. 97.8
SCHOLASTIC ACHIEVEMENTS
Awarded the Academic Excellence Award for good overall academic performance in the academic sessions 2014-15,2015-16, 2016-17
Secured AIR 386 in JEE-ADVANCED, 2014 and AIR 194 in JEE-MAINS, 2014 among 1.4 million registered applicants nationwide
Selected in KVPY Mentorship Scheme 2013 and also selected for the National Science Camp held at IISC Bangalore by DST, India
Qualified Regional Mathematics Olympiad(2013) with a State Rank 7 and got selected for Indian National Mathematics Olympiad
Qualified National Talent Search Examination (NTSE) in 2010 (organised by NCERT) among 0.3 million registered candidates
Selected for the Indian National Physics Olympiad (INPhO) in 2014 organised by Homi Bhabha Centre for Science Education
WORK EXPERIENCE AND KEY PROJECTS
Risk Analyst, American Express (Received PPO) (May’17-Jul’17)
Objective o Build and implement a statistical model to predict first party fraud events by corporates clients of the company
o Built an implementable solution using expenditure data comprising of 14 million transactions over last 12 months
Approach o Extracted data for their Decision Tree model, performed quality checks and tested different versions of the model
o Created opportunities to improve model’s performance through optimization of ML algorithm parameters
Impact o Pointed out improvement opportunities in existing method for creating the dependent variable for the model
Image Captioning (Aug’18-present)
o The project objective was to generate meaningful and semantically correct captions of images taken from MS COCO dataset
o Approached with breaking down images with Convolutional Neural Networks as the encoder for the image feature extraction task
o The encoding vectors extracted from the image were fed to RNNs (decoder) to generate meaningful sentences describing the images
Black-Scholes Option Pricing Model (Jan’18-Apr’18)
o Modelled the stock price movements as a Geometric Brownian Motion fitting normal distribution to historical stock returns data
o Applied the Black Scholes framework under ideal capital market assumptions for theoretical computation of CE option prices
o Predicted the trend and price for SBIN and SAIL options using underlying stock market data from NSE with ~90% accuracy
Robust MVDR Beam-forming using EM algorithm (Jan’18-Apr’18)
o Estimation of the time-frequency masks using probabilistic speech spectral model based on a Complex Gaussian Mixture Model
o Steering vector estimation from TF-bin probabilities for the signal, determined using the Expectation Maximization algorithm
o Designed MVDR beam-former for noise robust speech enhancement with 2.7 fold improvement in SDR against traditional algorithm
Robust Change Point Detection in Cognitive Radio Networks (SURGE’16) (May’16-Jul’16)
o Studied robust change point detection in cognitive radios, analysed the various change point detection paradigms and found out the
most efficient algorithm for the standard case of a single sensor taking its observations as scalar (Centralised Case)
o Extended the standard theoretical case discussed in the paper to the real life case of multiple sensors taking observations as vectors
o Compared the performance of possible algorithms for the case of multiple sensors with each sensor taking vector observations
(Distributed Case)and derived bounds on the performance parameters for the best performing (Majority Decision) algorithm
RELEVANT COURSES *ongoing
Machine Learning Techniques* Data Mining* Data Structures and Algorithms Linear Algebra
Stochastic Processes Probability and Statistics Speech Signal Processing Computational methods
POSITIONS OF RESPONSIBILITY
Head of Show Management, Techkriti'17 (CORE TEAM) (May’16-Apr’17)
o Managed a budget of approximately INR 6.5 million for the organisation of one of Asia's biggest technical festival- Techkriti
o Coordinated in the execution and planning logistics for a number of events at all levels in Techkriti'17 throughout the year
o Negotiated with various high level authorities to maximize the utilization of institute inputs for betterment of the festival
o Planned and executed Techkriti Open School Championship across 15 cities with a total of 10000 enrolled participants
o Ensured availability of various infrastructural and technical requirements of other team members for various events beforehand
TECHNICAL SKILLS
Languages: C,C++,Python
Tools: MATLAB,SQL, MS Office, LaTeX
