Mehant Kammakomati

I have graduated from the National Institute of Technology, Andhra Pradesh, with a Bachelor of Technology in Computer Science and Engineering. So far, I have worked with some of the leading industrial research facilities, including CERN Switzerland , IBM India Research Lab , and SCoRe Lab Sri Lanka. I have made my hands dirty in a diverse set of fields in Computer Science and Technology such as Machine Learning, Computer Vision, Software Engineering, and DevOps.

 /   /   /   /   /  Google Scholar  /  Stack Overflow  /  Internet

Timeline

  • [2022]   Joined IBM India Research Lab as Research Software Engineer.
  • [2022]   Became Kaggle Expert for datasets. Prepared ATM-I, ATMA-V, and COB datasets.
  • [2022]   Graduated from National Institute of Technology, Andhra Pradesh with B.Tech in CS.
  • [2022]   Completed SDE Internship at IBM India Software Lab.
  • [2022]   Became Google Summer of Code Mentor for Kubebot project under SCoRe Lab.
  • [2022]   BTech Thesis project: Anomaly Detection in ATM Vestibules using Three-stream Deep Learning Approach is accepted.
  • [2021]   Won JPMorgan Chase EMEA Code for Good Hackathon for building Helpify App.
  • [2021]   Scored 315 / 340 in GRE test and 100 / 120 in TOEFL test.
  • [2021]   Completed Internship at CERN openlab Switzerland as summer student and published a project report.
  • [2021]   Completed Internship at IBM Research as research software engineer and won Coding Champion Award.
  • [2021]   Became Google Summer of Code Mentor for Scan8 project under SCoRe Lab.
  • [2021]   Secured 6522 rank in Google Hashcode.
  • [2020]   Won Twitter Codechella International Hackathon category prize for building Tweepsume App.
  • [2020]   Finalist at Smart India Hackathon for building Antimicrobial Stewardship Platform.
  • [2020]   Completed Google Summer of Code as a student on project Bassa under SCoRe Lab.
  • [2020]   Became Google Code In Mentor.
  • [2020]   Worked as Mentee with Red Hat on project ansible bender.
  • [2020]   Won GNOME CE Challenge phase 1 for building project WelFOSS.
  • [2019]   Became Google Code In Mentor under SCoRe Lab.
  • [2019]   Completed Google Summer of Code as a student on project Bassa under SCoRe Lab.
  • [2019]   Built a cross-platform addictive game: UnHybrid that is ranked 54 / 322 global entries on GitHub GameOFF.
  • [2018]   Won Google Code In Grand Prize and visited Google MTV HQ office.
  • [2018]   Scored 970 / 1000 in 12th and ranked 15k in JEE Main test.
  • [2016]   School Topper in 10th standard with 96% (overall). Recipient of NRK Moorty Gold Medal and Silver Medals in Science, Computer Applications, and Telugu language.
  • [2015]   Served as Head Boy for the school and won Master Niraj Award .

Experience
irl

IBM Research (IRL) | Research Software Engineer


Research Areas: App Modernization, Hybrid Cloud, Code Analysis, Replatforming Tools
Tech Stack: Golang, Python, Docker, Kubernetes
Supervisor(s): Ashok Kumar, Dr. Padmanabha Venkatagiri Seshadri

Currently part of the Hybrid Cloud research team at IBM IRL, where we are building next-gen replatforming tools.

IBM ISL

IBM India Software Lab (ISL) | SDE (Intern)


Tech Stack: JS, React, Angular, Node JS, MongoDB, Carbon Design
Supervisor(s): Marianne Flahaut

Worked with AI Apps design team at IBM Canada Lab to develop Sponsor User Program (SUP) Portal.

irl

CERN Openlab (BE-ICS-FT) | SDE (Summer Student)


Tech Stack: Python, Docker, Databases, PostgreSQL, TimescaleDB
Supervisor(s): Rafal Kulaga, Dr. Anthony Hennessey

Worked on benchmarking tools and performance evaluation of TimescaleDB and PostgreSQL for storage of historical data from WinCC OA SCADA systems.

irl

IBM Research (IRL) | Research SDE (Intern) | Coding Champion Award


Research Areas: App Modernization, Hybrid Cloud, Code Analysis
Tech Stack: Golang, Python, Docker, Kubernetes
Supervisor(s): Ashok Kumar, Dr. Padmanabha Venkatagiri Seshadri

Reported optimal design choices for Netflix OSS transformation and decoupling EAR into EJB, WAR, and JAR to containerize and re-platform to Kubernetes.
Wrote transformer plugins natively(Golang) to analyze source artefacts (Netflix OSS and Rust) and generate IR and target platform artefacts.

irl

Google Summer of Code - SCoRe Lab | Student Developer


Tech Stack: Python, Golang, Bash, Kubernetes, Docker, Locust, Flask, MySQL, Prometheus, Grafana

(2020) Wrote Kubernetes scripts for monitor and alert functions of Prometheus
Wrote Kubernetes scripts to visualize cluster metrics using Grafana
Wrote client libraries in Python and Go with timeouts,retry, and rate-limiting

(2019) Wrote optimal dockerfiles for Bassa (Web, Server, Aria2c, and MySQL DB)
Wrote Kubernetes scripts for on-premise deployment over a single node cluster

Entrepreneurial
IBM ISL

Co-founded MergeURL.com (LIVE)

#2 product of the day on Product Hunt
1.55M pageviews and 4.84K 30 day active users.
Featured on Product Hunt, MongoDB Code Examples, student spotlights, and other blog and news articles.

Are you sick of posting multiple URLs? Here, you can create a tailored MergeURL hassle-free without any user registration.

Open Source Contributions

Major

IBM ISL IBM ISL IBM ISL IBM ISL

Minor

IBM ISL IBM ISL IBM ISL IBM ISL
Publications
cern report

Performance Evaluation of TimescaleDB for Storage of Historical Data from WinCC OA SCADA Systems


Mehant Kammakomati
Zenodo | CERN Openlab Summer Student Report
report

This project was completed in the scope of NextGen Archiver (NGA) for WinCC OA SCADA systems. The NGA is a new archiver for WinCC OA that uses a pluggable architecture to support multiple database technologies. This project was part of a wider effort to benchmark a range of database technologies to understand their limits in terms of functionality and performance in the context of CERN use cases. The benchmarking methodology involves producing realistic test data and performing write and read benchmarks on the database technologies under the test. The specific focus of this project was to perform ingestion benchmarking on TimescaleDB and PostgreSQL. We obtained an ingestion rate of 80K rows/second for TimescaleDB one-node and 150K rows/second for TimescaleDB two-node making them 2X and 3X higher than that of PostgreSQL which is around 40K rows/second. As the benchmark runs progressed we observed a considerable decline in the ingestion rate for PostgreSQL, but the ingest rate was stable for TimescaleDB. The work on the benchmarks will continue and focus on query performance and evaluation of different schema variants.

springer

Comparison of Machine Learning Algorithms for Hate and Offensive Speech Detection


Mehant Kammakomati, P. V. Tarun Kumar, K. Radhika
Evolutionary Computing and Mobile Sustainable Networks. Lecture Notes on Data Engineering and Communications Technologies (LNDECT). Springer Singapore
paper

Hate speech is not uncommon and is likely practiced almost on every networking platform. In recent times, due to exponential increase in Internet users and events such as the unprecedented pandemic and lockdown, it showed increased usage of social platforms for communicating thoughts, opinions, and ideas. Hate speech has a strong impact on people’s lives and is one of the reasons for suicidal events. There is certainly a strong need to make progress toward the mitigation of hate speech. Detection is the primary step to eradicate hate speech. In the following paper, the comparative analysis of different machine learning algorithms to detect hate speech was shown. Data from the Twitter social platform was considered. From the analysis, it was found that the long short-term memory method is a highly performant machine learning algorithm.

mergeurl

MergeURL: An Effective URL Merging and Shortening Service


Mehant Kammakomati, Sai Vittal Battula
International Journal of Computer Science and Mobile Computing
paper

Multiple URLs belonging to the same purpose can become cumbersome to display, share, and handle. In this paper we describe MergeURL, an n-tier application to merge and shorten multiple URLs instantly overcoming the barriers of authentication and registration process while not compromising with security and resistance to data redundancy.

Selected Projects

Anomaly Detection in ATM Vestibules using Three-stream Deep Learning Approach

Code | Thesis Report
BTech Thesis | Advisor: Dr. Karthick Seshadri

Anomalies are abnormal events that deviate from typical behavioral patterns. Video anomaly detection is the problem of identifying anomalies in video feeds. ATM vestibules are one of the critical places where such anomalies must be detected. The problem lies around how we represent a video and further perform analysis on it to predict an anomaly. Another problem is the unavailability of data for this task specific to anomalies that can happen in ATM vestibules. To tackle these, this report presents the usage of a three-stream deep learning architecture and introduces two novel datasets: box annotated image and temporal annotated video dataset. The three streams correspond to contextual, spatial and motion information extracted from the video feed. It is first-of-its-kind attempt to leverage box annotated image dataset for fine tuning the object detection model to detect ATM class and temporal annotated video dataset to train the model to detect anomalies in video feeds featuring ATM vestibule. The presented work achieves a recall score of 0.94, and false positive rate of 0.13.

Selected Positions Of Responsibility & Extracurricular Activities
mentor

Google Summer of Code Mentor for 2021 & 2022

Mentored Projects: Kubebot, Scan8 (Full Stack), Scan8 (DevOps)

Mentoring involves reviewing code, suggesting architectures, discussions, and mentee evaluations.
mentor

Google Code In Mentor for 2019 & 2020

Mentored and Reviewed code for 500+ Students

Mentoring involves assigning tasks, reviewing code & design, evaluating subsmissions, and guidance to participating students.
mentor

Started NIT Andhra Pradesh Open Source wing

Hosts unofficial community projects.

 /   /   /   /   /  Google Scholar  /  Stack Overflow  /  Internet