Ankit CV
Ankit CV
Ankit CV
PROFESSIONAL EXPERIENCE
Trainee Consultant, Mastercard Advisors, India June 2019 – present
Completed Delivered credit-card loyalty solutions and target audience recommendations for a Spanish bank
Projects Designed data-driven marketing campaigns for merchants based on customer transaction analysis
Ongoing Developing an audience recommendation engine for merchants based on customer transaction data
Projects Utilising user data and metadata; implementing Alternating Least Squares algorithm – Hadoop, Spark
2/26 in Idea Competition; Idea: Open-banking to curb price-volatility on e-commerce platforms
Achievements
Awarded USD 7500 (team of 5) to develop idea into an end-to-end marketable product by March, 2020
Patents Automatic checkout on ecommerce-platforms at consumer’s target price; under internal review
JOURNAL PUBLICATIONS
1. A unified deep-learning model for classifying the cross-country skiing techniques using wearable gyroscope sensors
Sensors 2018, 18, 3819
Authors: Ankit, A; Jang, J.; Kim, J.; Jang, Y.J.; Kim, H.Y.; Kim, J.H.; Xiong, S, Citations Count: 5
2. Economic design of variables acceptance sampling plan with rectifying inspection based on advanced capability index
Authors: Ankit, A.; Song, W.T.; Manuscript under preparation
RESEARCH EXPERIENCES
1. Bachelor’s Thesis, IIT Roorkee July 2018 – May 2019
Supervisor: Prof. Akshay Dvivedi, IIT Roorkee
Project Title: Design of multiple dependent-state variables-based acceptance sampling plans with rectifying inspection
• Utilized Cpmk capability index to design novel sampling plans; considered process yield and quality loss simultaneously
• Framed the problem as multivariate non-linear optimization; solved using optimization solver on MATLAB
• ↓ quality cost & sample size by 8%; ↓ confidence interval size for µprocess by 25% w.r.t previous state of the art plans
2. Korea Advanced Institute of Science and Technology, South Korea July 2018 – October 2018
Supervisor: Prof. Shuping Xiong, Human Factors and Ergonomics Lab
Project Title: Developing a classification model for cross-country skiing techniques to train amateur skiers
• Framed the problem as a multivariate time-series sequence classification task using signals from wearable sensors
• Performed signal processing on MATLAB; developed a CNN-LSTM model on Python – Keras & Tensorflow
• ↑ accuracy by 12% (79%→91%) and ↓ the no. of sensors for data collection from 21→5; new state of the art
• Designed economic model of variables single-state sampling plans with rectifying inspection for multistage SCs
• Derived cumulative distribution function of Cpmk index; solved constrained non-linear optimization on MATLAB
• ↓ quality cost & sample size by 7.8%; ↓ confidence interval size for µprocess by 25% w.r.t previous state of the art
C-1517, Sushant Lok, Gurgaon, India - 122002
Ankit ankit@mastercard.com, ankitvyas.me@gmail.com
+91-9897426734, +91-8929880430
4. Independent Research Work October 2018 – Nov. 2018
Project Title: Sparsification study of neural networks to make them deployable on computationally weak systems
• Pruned k% weights and neurons of hidden layers based on L1 and L2 norms of weights; retrained the sparse network
• Results before retraining: [Sparsity, test-time speedup, loss in accuracy] = [50%, 2.40x, 6.86%]
• Results after retraining: [Sparsity, test-time speedup, loss in accuracy] = [90%, 7.06x, 1.94%]
INDUSTRIAL EXPERIENCES
1. Jindal Stainless Limited (JSL) May 2017 – July 2017
Project: Benchmarking and reduction of turnaround time (TAT) for outbound logistics within the plant premises
• Bifurcated & benchmarked TAT into value-added (15%) & non-value added (85%) time; proposed TAT ↓ by 30%
• Successfully conducted trial run of opening checkout in night shift which decreased TAT by 30 mins/truck
2. ITC Nov. 2016 – Jan. 2017
Projects: i) Optimum design of shop floor plan ii) Productivity improvements of hot foil stamping machine
• Designed & implemented an improved shop floor plan based on work and time study; ↓ worker movements
• Decreased downtime of hot foil stamping machine by 8 min/hr, increased output by 400 sheets/hr, approx. 15%
COMPUTER SKILLS
Programming Tools: Python (Keras, TensorFlow, scikit-learn, Numpy, Pandas), MATLAB, C++, LaTeX
DBMS: SQL, Proc SQL, SaaS, Apache Hadoop Softwares: Xsens MVN, ANSYS Fluent, SolidWorks
RELEVANT COURSEWORK
Graduate Level Electives: Product and Process Optimisation, Product design and Development
Supply Chain Management, Design for Manufacturability
Undergraduate Coursework: Probability and Statistics, Linear Algebra and Multivariate Calculus, Operations Research
Operations Management, Economics and Engineering Economics, Quality Management