Ercag P
- Research Program Mentor
PhD at Bilkent University
Expertise
Data Science, Physics & Statistics
Bio
Hi! I'm Erçağ (pronounced like Archer). I'm a senior scientist and PENN Data Science fellow at the University of Pennsylvania. I hold a PhD in Physics and am currently pursuing a MSE degree in Data Science at UPENN. My research focuses on understanding the physics behind bacterial swimming and its role in the assembly of large microbial communities. Specifically, I study bacterial locomotion and how cells navigate through complex environments. My work involves conducting experiments, analyzing swimming statistics, and developing machine learning models to elucidate microbial movement patterns. Beyond my scientific pursuits, I am also interested in building tools to analyze time series data using state-of-the-art LLMs and transformers. Another research topic I am invested is how to brew perfect espresso informed by physics and statistical learning models. What is the kitchen science behind brewing delicious espresso with the least amount of coffee beans? Exciting stuff! When not in the lab or glued to a computer screen, I enjoy boxing, reading Russian classics and watching arthouse movies. I am quite passionate about curating a collection of cult art house movies, and Criterion Collection movies are my favorites!Project ideas
Harnessing generative AI in forecasting stock market trend
The generative AI, specifically large language models (LLMs) has taken the world by storm over the last few years. From simple tasks of note-taking to writing a well-developed text, LLMs such as Chat-GPT are already becoming vital tools in our daily lives. Despite its main area of usage, generative AI can also be harnessed to build forecasting models to predict future trends. In this project, we will leverage transformers, the basis of the attention mechanism, to set up a framework for public time-series forecasting. Once the proof-of-concept phase is complete, we will analyze the available stock market data and stress-test the developed AI tool. A potential outcome of this fruitful project will be to obtain a solid framework and toolkit for future forecasting analyses and create a ready-to-use forecasting app. This is an advanced-level project.
Brewing perfect espresso using machine learning tools
Brewing great espresso is not an easy task. Multiple components are involved from Physics of fluids to Kitchen Science that one experiences every day. In this project, we aim to tackle the problem of brewing the perfect espresso with a more holistic approach. In particular, we will develop machine learning models on the brewing data with features varying from coffee weight to extracted volume. A significant part of the part will involve some fun experiments in which students can explore the parameter space in her kitchen. This is a beginner-level project.
Classification of bacterial motility using AI tools and machine learning
The motility of bacteria is known to play a significant role in sustaining eco-diversity and microbial coexistence. Bacterial species often exhibit a diverse set of swimming phenotypes that help to leverage dispersal in the face of competition for food and space. In this project, we will develop an AI-based tool to detect emerging patterns in bacterial trajectories. Based on the detected swimming phenotype, the computational tool will facilitate the classification of microbial species and reveal their taxonomy. The developed tool will help researchers to identify the class of motile microbes tracked under a microscope. This is an advanced-level project.