Download CV

Education

Indian Institute of Science, Bangalore, India
MTech (Reseach) - Department of Computational and Data Sciences (CDS), (July 2017 - present)

• Advisor: Dr Partha Talukdar
• Affiliation: Machine and Language Learning (MALL) Lab

Birla Institute of Technology and Science, Pilani, India
B.E. (hons.) Electrical and Electronics Engineering, (August 2012 - June 2016)

• Thesis Topic: "Asynchronous Distributed Stochastic Convex Optimization in MPI"
• Advisors: Dr Partha Talukdar (IISc, Bangalore), Dr Ashish Mishra (BITS Pilani)

Honours and Awards

Recipient of the Intuit Research Fellowship (2017 - present).\\

Awarded a $1000 Travel Scholarship to attend the International Conference on Empirical Methods in NLP (EMNLP 2017) at Cohenhagen, Denmark, 2017.

'Top 10' finish in the famed Allen Ai Science Challenge on Kaggle, 2016.

Invited to attend the 29th Annual AIAA/USU Conference on Small Satellites at Utah, USA on a $2400 Travel Award, 2015.

Publications

2018

Chandrahas, Aditya Sharma, Partha Talukdar. "Towards Understanding the Geometry of Knowledge Graph Embeddings." 56th Annual Meeting of the Association for Computational Linguistics (ACL 2018), Melbourne, Australia.

2017

Aditya Sharma, Zarana Parekh, Partha Talukdar. "Speeding up Reinforcement Learning-based Information Extraction Training using Asynchronous Methods." International Conference on Empirical Methods in NLP (EMNLP 2017), Cohenhagen, Denmark .

2016

Aditya Sharma, Pranet Verma, Ashish Mishra, K. Solomon Raju. ”A Modular Approach to Random Task Graph Generation.” Indian Journal of Science and Technology, February 2016.

Professional Experience

Intuit IDC, Bangalore, India
Tech Futures Team
Summer Research Intern, May, 2019 - July, 2019

• Built an unsupervised extractive query focused summarization system to create targeted summaries from Google News articles.

Intuit IDC, Bangalore, India
Innovation and Advanced Technologies Team
Summer Research Intern, June, 2018 - July, 2018

• Worked on building an adversarial reinforcement learning based conversational agent for automating Intuit's Customer Care using TensorFlow.

Indian Institute of Science, Bangalore, India
MALL Lab, Department of Computational and Data Sciences (CDS)
Research Assistant, August, 2016 - July, 2017

• Worked with Dr Partha Talukdar on applications of Deep Reinforcement Learning in Natural Language Processing. Specifically, worked on automating parts of the Knowledge Graph generation pipeline using intelligent agents.

CERN, (European Organization for Nuclear Research), Geneva, Switzerland
Google Summer of Code Student, April, 2016 - August, 2016

• Added support for compressed Deep Neural Networks along the lines of HashedNets to TMVA (Toolkit for Multivariate Data Analysis with ROOT), a popular open source software for data analysis used by the High Energy Physics community, maintained by the Software Team at CERN.
• Code developed as part of the project is available here.

Indian Institute of Science, Bangalore, India
MALL Lab, Department of Computational and Data Sciences (CDS)
Undergraduate Thesis Intern, December, 2015 - April, 2016
Summer Research Intern, May, 2015 - July, 2015

• Implemented distributed Representation Learning models for NLP on the Cray XC40 supercomputer SahasraT, the fastest supercomputer in India, using MPI-OpenMP.
• Worked on asynchronous parallel implementation of convex optimization algorithm Stochastic Gradient Descent, along the lines of Downpour SGD.
• Developed a Parameter Server infrastructure on Cray XC40, from scratch, in MPI.

Centre for Development of Imaging Technology, Trivandrum, India
Research Intern, May, 2014 - July, 2014

• Implemented a lossless version of the JPEG2000 image compression algorithm in MATLAB, which was used for medical imaging applications by C-DIT.

CSIR-CEERI, Pilani
Student Research Assistant, August, 2014 - May, 2015

• Worked on software for an FPGA based fingerprint sensor.

Projects

BITS Pilani Student Satellite Team
Founder and Team Leader, November, 2013 - October, 2015

• Lead a team of 20 students to build a low cost {\bf hyperspectral imaging 3U CubeSat} weighing less than 10 kg.
• Engineered an FPGA based System-on-Chip low power, high performance data processor and architected the flight software for the satellite from ground up.