Research & Projects

Interested in the latest technologies but don't know where to start? What is the To-Do list to kickstart the transition to a career in tech? Is it the same for everyone? Think methodically. Know your “Why” behind career transition to tech. Pick your choices. Recognize the gaps. Start working towards them. Read the research paper to build your tailored kickstart To-Do list to leap to a career in tech.

Master_Presentation.pptx

Aegis : Privacy on Social Media

Master's project

  • Built a AI Cyborg (Aegis) which aims to operate at client's site to prevent location privacy breach.

  • Conceptualized and programmed a novel obfuscation privacy model such that adversary confidence is reduced by 59.7%.

  • Tested and optimized the model for 6.86 million tweets and 360 user's profile.


Differential Privacy

Cryptography Project

  • In-detailed researched the details of differential privacy.

  • Analysed various use cases of application of differential privacy.

  • Critiqued the literature available in the domain.

ish+pooja_final_midnight

Ramsey Number : Using all the clouds

Advanced cloud computing

  • Implemented 5 private + public clouds to solve a computationally intensive problem by doing automatic job scheduling and job slicing.

  • Architected an overlay on 50+ commodity instances to make maintanence and control easy and flexible by using terminal multiplexing software.

Monday_before_ppt_final_first

Technology Stack

Ruby on Rails, Tsung , Git, Travis CI

Boomerang : find people in the same mood 🔥!

Scalable Internet Services

• Boomerang is an application that connects people together. People would love and appreciate real, safe and simple communications in person with the help of our application which matches people based on thier mood.

• Identified bottlenecks for scalability in load balancing and SQL statement using Tsung load testing.

• Re-designed the system using pagination, caching and scaling (horizontal & vertical) techniques to improve performance by 33.3%.

RydeCoin : Peer to Peer ride Sharing on Blockchain

Global Data Management

  • Developed a ride sharing application on top of a private blockchain to create a trusted but decentralized system such that we de ne a contract between the provider and the consumer without any middle-man.

  • Data dissemination protocol optimized blockchain network performance and scalability by splitting workload across all nodes by 30%

rydeCoin [Autosaved].pptx

  • Introduced Chord for broadcasting the commited blocks in the blockchain. Investigated the placement of nodes and their reachability on the over-all performance of our system in terms of commit latency and throughput.


sunday_final.pptx

Rust, C, Java Profiling

Runtime Systems

  • Build an automated profile system for profiling languages like C, Rust and Java

  • Learnt and analyses there relative perfornace and internal working in memory management.

KOS : An operating system for a simulated MIPS machine

Operating System

  • Developed and Coded my own version of malloc() for dynamic memory allocaion. Handled memory fragmentation.

  • Wrote 8 system calls (like fork, wait, sbrk) and constructed the memory management system

Distributed Movie Ticketing System

Advanced Distributed Systems

  • Designed a online ticket system for 5 kiosks such that mutual exclusion is ensured.

  • Implemented multithreading and heartbeat system to execute PAXOS consensus algorithms.

  • Rebuilt app to tolerate site failures and network partitioning by collaborating a global snapshot algorithm.

  • Simulated state machines in Python by establishing reliable TCP/IP connections.

Social Network Analysis for Advertisements

Dynamic Network Science

  • Constructed an application for advertising a local event deciding who all to give information about the event so that they cover maximum people in a community.

  • Engineered problem into a wider problem of choosing a committee in a multi-attributed network which ensures maximum diversification as well as coverage.

  • Quntified the problem with an algorithm built on greedy algorithm leading to improving the runtime by 60%.

  • Piloted the model on a social network data set in Python. Reached 30% coverage in the data set.

Technology Stack

Python , SQL , Git