In this role, I specialized in matching algorithms and bias mitigation, contributing to cutting-edge solutions that enhance diversity and fairness in AI-driven processes:

• Developed internal dashboards with Python, Flask, React and Amazon Athena SQL, that analyze bias of AI/ML systems ranking 10M+ applicants for 100K+ jobs by monitoring metrics across protected categories (e.g. race and gender) to identify bias concerns, contributing to enhanced fairness and validating legal compliance of ranking systems.

• Owned the design and development of automated resume perturbation testing infrastructure capable of generating test datasets (250+ applicant resumes, 100+ job requisitions) from customer data and measuring sensitivity of ranking systems across 5+ perturbation categories (e.g. gender, disability), reducing the time to execute the tests from 1 week to only 1 day.

• Led a 4-person team to develop a tool identifying and replacing discriminatory language in job descriptions, fostering more inclusive job descriptions and reducing bias in recruitment.

• Integrated OpenAI Large Language Models (LLMs) into Eightfold platform to generate job descriptions tailored to a given list of skills and experiences required for jobs, streamlining the process of writing a job description to 10 seconds and improving quality of job descriptions.

• Authored a white paper and an educational blog post on the topic of bias in AI technology systems, disseminating knowledge and best practices to a wider audience of 1000+ readers, and promoting transparency and responsible AI development within the industry.

• Improved the performance of ranking systems in non-English languages by enabling translation of certain keywords such as applicant's skills and experiences in downstream Natural Language Processing (NLP) tasks utilizing Google Translate API, expanding the platform's global reach and effectiveness in 5 languages.

I researched computer vision and machine learning technologies focused on vehicle tracking at traffic intersections for an MCity project at University of Michigan funded by Denso under supervision of research scientist Brent Griffin. Our research focused on the following content:

• A multi-camera vehicle tracking system made of Faster R-CNN, SORT and fully connected neural networks (PyTorch) that can estimate GPS position of vehicles observed by different traffic cameras with a median error of 0.15 meters

• Python tools for a virtual environment that generates randomly shaped vehicles following random trajectories and corresponding observations of these vehicles in multiple camera views to train and evaluate our multi-camera vehicle tracking system

• Genetic Algorithm-based method to estimate intrinsic and extrinsic parameters for camera models that visualize real-life cameras in our virtual environment given a camera image, pixel values of selected calibration points and their corresponding GPS coordinates

Worked as a Student Software Engineer in a UM Multidisciplinary Design Program project with Little Caesars Enterprises to build a Quality Assurance software for Little Caesars' pizzas:

• Built Quality Assurance tool for Little Caesars' pizzas utilizing Machine Learning and Computer Vision tools in collaboration with 5 other project-mates to improve pizza quality and decrease QA process to under 6 seconds.

• Managed database of 5000+ objects in Azure Blob Storage with Python programs to ensure database is synchronized with the latest local versions and  is available for use

• Configured Virtual Machines on Azure to build framework for cloud-based software system that provides 10x faster operation of Machine Learning models

Researched security vulnerabilities in computer systems of vehicles in the Real Time Computing Lab at University of Michigan under Professor Kang G. Shin and Ph.D candidate Mert D. Pese:

Analyzed vulnerabilities in Passive Keyless Entry Systems (PKES) of vehicles with Yard Stick One and RTL-SDR to demonstrate how PKES can be compromised with denial of body attacks

I worked as a student researcher in the Assistive Vision Team for the Multidisciplinary Design Program at the University of Michigan for two projects:

Surgical Instrument Tracking:

With the help of my teammates, I developed software that analyze surgical instruments in eye surgery videos with hopes of understanding what surgery skills do surgeons need to perform successfuly surgeries. Our software tracked the surgical instruments by measuring the position and orientation using computer vision tools and machine learning algorithms such as OpenCV, RANSAC, and AprilTags. 

Smart Cane for Visually Impaired

Our team designed a smart cane that can help visually impaired in daily life by notifying them of objects in their surroundings. We created a system of ultrasonic sensors, vibration motor discs, Arduino Uno, and a white cane to build the Smart Cane. The smart cane was capable of measuring the distance of the objects in the surroundings and warning user with different vibrations depending on the distance of the objects.

I interned as a software product manager for the SmartAisle Team at The Mars Agency in the summer of 2019. Our team worked on a scalable, digital in-store assistant that provides shoppers with expert product selection guidance, information, and education at the shelf.

As the software product manager intern, I was responsible of defining why, how, and what product features our software engineers needed to develop. I led the team of 5 engineers to build a centralized system for log entries coming from different components of our product system, designed a performance analysis tool to triage errors from log entries and crafted an email notification system for maintainence engineers to be aware of system failures. 

I also developed internal tools that helped the business team to measure potential value of clients and manage relationships with clients.

I worked as a translator for WafaGames, a video game company based in Beijing, China focused on mobile strategy and SLG games, in the summer of 2018. I translated one of their games, Swords of Glory from English to Turkish and German languages. With my help the game had local language support and appealed to the Turkish and German gamers better. 

At WafaGames I met Kathy Gong, the CEO and co-founder of Wafagames and an amazing entrepeneur who was once the youngest national chess champion in China. As she was my supervisor, I had the chance to learn many things from her on how to become successful in the video game industry and as a startup. 

Furthermore, my experiences in China and my interactions with the local people astonished me. I got inspired to learn more about the Chinese culture and learn Chinese (Mandarin) language after this experience.