About Me
I am currently working in Oracle Financial Services Software Limited at Bangalore in R&D department as a developer in Blockchain team. Here, we develop solutions using Hyperledger Blockchain and explore other Blockchains i.e. Ethereum and Corda.
Oracle Financial Services Software: (Application Developer 2)
- Working on projects where my role involves hands-on experience in Java, Blockchain (Hyperledger Fabric Chaincode), goLang, ActieMQ, Springboot, WebServices, Jenkins (CI-CD), Apache Camel, Docker and coordinating with multiple teams and mentoring new hires.
IIT Madras (Masters)
- Worked on research problems in the area of Big Data, Machine Learning, Graphs and Social Network Analysis. My work involved hands- on experience in Java, Apache Hadoop, Apache Spark, GoFFish, Graph algorithms, HDFS, Maven, Docker, Cloud and Virtualizations, Cassandra, MATLAB
- Published a research paper in an International conference which focuses on detection of cyber-communities in social networks, recommendations based on the interest group, and estimating hidden features in a social network
Publications:
-
DCEIL: Distributed Community Detection with the CEIL Score
High Performance Computing and Communications (HPCC), 2017
Akash Jain, Rupesh Nasre, Balaraman Ravindran
Community detection in complex networks has a wide range of applications such as detection of cyber-communities in social networks, recommendations based on the interest group, and estimating hidden features in a social network. In distributed frameworks, the primary focus has been scalability. However, the accuracy of the algorithm's output is also critical. We propose the first distributed community detection algorithm based on the state-of-the-art CEIL scoring function. Our algorithm, named DCEIL, is fast, scalable and maintains the quality of communities. DCEIL outperforms the existing state-of-the-art distributed Louvain algorithm by 180% on an average in Normalized Mutual Information (NMI) Index and 6.61% on an average in Jaccard Index metrics. DCEIL completes execution for 1 billion edges within 112 minutes and outperforms state-of-the-art distributed Louvain algorithm by 4.3 x. DCEIL critically exploits three novel heuristics which address the existing issues with distributed community detection algorithms that have the hierarchical structure of CEIL or Louvain methods. Further, our proposed heuristics are generic as well as efficient, and we illustrate their efficacy by enhancing the accuracy of distributed Louvain algorithm by 22.91% on an average in Jaccard Index, and the average execution time by 1.68 x over popular datasets.
Previous Work Experience
I have worked in Computer Sciences Corporation India Pvt. Ltd. from July 2012 - September 2014 as a Associate Professional: Product Developer in new business Accelerator team. Here I assisted in enhancing a CSC’s in-house product nbA by completing the development and deployment of several modules. I have helped in setting up a local development environment to expedite debugging and development, thereby reducing the turn around timefor new projects. Also, Cleared the Product Knowledge Framework Examination. I provided support and fixed bugs also mentored and taught new resources to work collaboratly.
Academic Details
I have completed my M.S. from Indian Institute of Technology, Madras in Computer Science and Engineering in 2018. I worked under Dr. Balaraman Ravindran and co-guided by Dr. Rupesh Nasre at IIT Madras. I joined IIT Madras in July 2015. My area of research is Machine Learning and Distributed Graph Processing Frameworks for Complex Network Analysis. I am working on community detection algorithms for distributed graph processing frameworks (i.e. GraphX, GofFish) to figure out structural patterns in dynamic graphs. I have completed my B.E. from Institute of Engineering and Technology, Devi Ahilya Vishwavidyalaya, Indore in Computer Science and Engineering in 2012.
Contact:
akki [at] cse[dot]iitm[dot]ac[dot]in