- Developed GIT-PIR, a private information retrieval solution, using Golang, C/C++ and Docker to integrate linearly homomorphic encryption with Git, enabling users to privately clone Git repositories without exposing their requests to the server.- Extended Git's source code to introduce a new data transfer protocol for GIT-PIR and demonstrated its scalability in a simulated Git hosting environment and its compatibility with code hosting platforms like GitLab and GitHub.- Analyzed 23.3 million CircleCI builds to identify and mitigate recurrent workload congestions in CI pipelines, optimizing executions through predictive scheduling and achieving a 23% reduction in average build times.Technologies Used: Git, GoLang, C/C++

Bykea is an on-demand transportation and logistics platform offering ride-hailing, food and grocery delivery, parcel delivery, and bill payment services, primarily in congested urban areas.Bykea Food Delivery:- Led the development and successful launch of a multi-store functionality, enabling efficient inventory and order management across multiple stores.- Managed a team of 4 developers in designing and developing a scalable food delivery microservice.- Collaborated with product managers to prioritize feature development based on user feedback, resulting in a 20% increase in customer platform engagement.- Conducted peer code reviews and provided constructive feedback to ensure adherence to coding standards.Technologies Used: Typescript, Node.js, Express.js, PostgreSQL, Redis, Dockers, AWS EC2

Dubizzle Group owns and operates bespoke classifieds portals in the emerging markets of online portals, primarily in the MENA, South Asia, and Southeast Asia regions. Bayut CRM:- Collaborated with the Bayut UAE team to provide a unified CRM experience across multiple platforms in 18 countries.- Identified performance bottlenecks in the legacy codebase by conducting extensive benchmarking tests on existing APIs,pinpointing the areas for improvement and resolving them by optimizing slower queries.- Played a pivotal role in requirements gathering and validation for the CRM application, ensuring the creation of clear,unambiguous technical specifications and establishing a consistent feedback loop.- Successfully executed a seamless data migration from Salesforce to Bayut CRM, ensuring 99.9% accuracy in transferring over 500,000 client records.- Conducted a POC on Laravel unit testing and provided training to the legacy team, resulting in a 15% improvement in overall test coverage.Technologies Used: PHP, Laravel, MySQL, Redis, Vue.js, AWS EC2

Caremerge is a healthcare technology company specializing in care coordination and communication solutions, improving senior living through streamlined communication, family engagement and efficient care management.Caremerge Voice (Alexa Skill):[AWS (Lambda, DynamoDB, SQS, S3, KMS) - NodeJS]- Worked with Amazon Engineering and Solutions Team to successfully publish Alexa’s first invocation-free skill, Caremerge Voice, for senior living.- Responsible for database design, code architecture, implementation and release of new features and enhancements of the skill.- Developed new Intents and Interaction Models for Amazon Alexa Skill.- Introduced encoded homophone representation for similarity matching to solve challenges faced by speech-to-text synthesis.- Designed and implemented metric reporting of individual intents to track interaction flow.- Implemented progressive responses to bridge the gap between API delays, creating a lag-free interaction for the end user.- Developed a Unit Testing and Code Coverage Framework from scratch using Bespoken, MochaJS, ChaiJS, SinonJS and Istanbul.Caremerge AI:[Python - Tensorflow - Scikitlearn - Pandas - Numpy - Heroku - Flash]- Developed and deployed a machine learning model to validate patients' medication data, achieving an accuracy of 97% on the test dataset.- Created a multi-label classifier using Tf-IDF vectorization to predict the medication type and frequency of a patient's medication prescriptions.- Employed feature selection and used non-linear dimensionality reduction to cater to highly imbalanced classes.- Revised and re-trained model using SGD and Kernel Approximation.- Deployed the model on Heroku and created an API for the web application.Caremerge Web Application:[NodeJS - HTML/CSS - AngularJS | Redis - RabbitMQ - Sequalize - MySQL - Postgres | Mocha - Chai - Istanbul | Dockers]- Added new features and enhancements to the product.- Created admin dashboard and wrote APIs for Caremerge Voice.

As part of our undergraduate thesis, under the supervision of Dr. Fareed Zaffar, and Dr. Rashid Tahir, (PhD UIUC) my research team and I have worked on a tool called "cloudGuard", that can detect malware attacks in real-time, and prevent abuse, despite an active adversary trying to bypass defenses.Our system aims to safe-guard cross-VM and cloud infrastructure and primarily focuses on detecting the signature of Hardware Based, Side Channel Attacks. The system employs hardware-assisted profiling coupled with machine learning techniques to create discernible signatures for various recent side channel attacks including Rowhammer, ZombieLoad, Flush+Reload, Prime+Probe, and Flush+Flush in real-time, with negligible overhead, despite an active adversary trying to bypass defenses, and can accurately detect these with negligible overhead. The results so far have been really exciting (>98 % detection accuracy with noise incorporation) and empirically demonstrate the uniqueness of hardware attack behavior. We plan to submit the work to a top security conference this year.

CS 452 - Computer Graphics- Delivered lectures- Conducted regular tutorials- Kept weekly office hours- Graded assignments

Implemented a convolutional neural network (ResNet-152) on satellite imagery to achieve a 78% accuracy in identifying bonded labor areas around brick kilns in southwest Asia.Github: https://github.com/mkarsalan/BrickKiln-Prediction-Using-Satellite-Imagery

Developed an Android application that uses point triangulation, bilinear interpolation, and real-time data to generate heatmaps for weather parameters like humidity, temperature, and AQI, visualizing pollution effects across the city.

- Worked for 8 weeks as an engineering intern. - Automated the web application using Selenium and Protractor.- Maintained Jenkins processes.

Constructed architecturally accurate models using point clouds and developed a VR tour in Unity3D incorporating site descriptions, ambient sounds and interactive elements, offering users a multifaceted exploration of the historical landmark.Technologies Used: Unity3D - C#Github: https://github.com/mkarsalan/Derawar-Fort-Unity3D