PROJECT FEED

The project analyzes a large federal program that subsidizes rural internet (broadband) projects. Where did the money go? Why are certain projects more expensive than others? What are the characteristics of the winning bids?

With a 4-letter code our genomes instruct when, where and to what extent each gene should be expresses. This underlies the acquisition of cellular identify during embryonic development; A cell at a specific position in the embryo, at a specific time in development expresses a subset of genes at precise levels, dictating a distinct repertoire of proteins that shape cellular morphology and behavior. Correspondingly, misregulation of gene expression often underlies developmental disorders. In the lab we work to decode these regulatory instructions from the genome.

We are a young vibrant lab looking for undergraduate students to join Dr. Brian Gills research team. Our lab's interest is in brain tumor-neuron interactions and how they contribute to tumor associated epilepsy and tumor progression. We are investigating if mTORC inhibition can reduce tumor associated seizures and reduce neuron-driven tumor progression. We use a variety of novel techniques to investigate these goals including ex vivo patch clamp recordings of mouse glioma models; ex vivo human tissue microelectrode recordings; and in vivo calcium imaging of peritumoral excitatory neurons in mouse glioma models. Projects are available for students interested in computational analysis.

AEQUITAS Lab is a research lab situated in Teachers College, conducting methodological and empirical research in two directions: 1) Understanding sources, patterns, and consequences of existing educational and social inequalities with data science and AI; 2) Interrogating the inherent equitability and extrinsic equity implications of data science and AI in education. The student will participate in one or more ongoing research projects and work with doctoral students on literature review, data analysis, writing and reporting, among other tasks.

The School Board and Youth Engagement Lab at Teachers College, Columbia University seeks a highly skilled and passionate UI/UX Designer and Researcher to join our team. Their role will be to support the development of Outreach, a civic engagement and deliberation tool designed to empower action and drive decision-making. This individual will be responsible for converting political science research-based goals for user engagement into intuitive, accessible, and impactful user interfaces and experiences. They will also be responsible for supporting user research by integrating data collection processes into the tool’s design.

Scientists at Columbia have recently developed a flexible new modeling suite called TokaMaker, which is able to reconstruct equilibria, assess stability and optimize time-dependent control schemes for tokamak plasmas. This will be an essential tool in the design and preparation of future fusion energy systems. In this project, we will develop a GUI interface for the new code, focusing on flexibility in use and advances in physics understanding. The GUI will be made publicly available to the fusion community, ensuring a broad and lasting impact for the code. Applicants should have a strong coding background, preferably in python.

The tokamak, a leading device design for magnetic confinement fusion energy reactors, performs well due to the capture of charged plasma particles on its toroidal flux surfaces. However, it is possible for the nice concentric flux surfaces to “tear” and open up “magnetics islands” that cause radial transport of particles and heat [YouTube Lecture on Islands, Poli IPP Thesis 2012 (Ch 1-2.1)]. Asymmetries as small as δB/B0 ~ 1e-4 (i.e. a perturbation “error field” 10,000 times smaller than the primary tokamak magnetic field) can drive a plasma instability. This project will look at a database of experiments that did this intentionally, and use that data to determine what level of asymmetry would cause tearing in a future reactor so operators might avoid it or force it as they please. Specifically, this project will concentrate on Machine Learning (ML) approaches for predicting the RMP thresholds. A second objective of the work will be to identify the most impactful new experimental observations that could be obtained within the available operating space. The ML based approach will require an investigation of the relative importance of different plasma parameters in determining this threshold, which will in turn guide what new experiments should be performed to improve our understanding.

The end-to-end performance acceleration of data center applications requires a large number of distributed accelerators as these applications exhibit wide diversity and operational complexity. Unfortunately, today’s systems community lacks the infrastructure needed to explore the large design space of distributed accelerators. To close this gap, in this project, we aim to build an accelerator-centric data center simulator. Our simulator will take dependency graphs of distributed applications and high-level descriptions of accelerator algorithms as inputs and investigate the design trade-offs of distributed accelerators. We will validate the effectiveness of our simulator with a key case study of where to place any given accelerator for data center applications, on a die, on a chiplet, on a CXL-attached device, or on a network interface card. We will release our research artifacts with flexible open-source licenses to better facilitate academic and industry research on accelerators for data center applications.

Harnessing the power of the endogenous immune system has emerged as a powerful strategy for disease prevention and treatment, spanning traditional vaccines to modern cancer immunotherapies. The Yousefpour Lab focuses on developing next-generation RNA- and protein-based vaccines and immunotherapies by integrating synthetic biology, protein engineering, and biomaterial science. We have multiple positions available. Students will contribute to the design and development of RNA-and protein-based vaccines, including optimizing RNA sequences, studying immune response mechanisms, and exploring innovative vaccine formulations. The projects will also involve broader work on RNA-based therapeutics, such as engineering synthetic biology platforms and biomolecular drug delivery strategies. Through these projects, students will gain hands-on experience in molecular biology techniques, in vitro cell culture, and characterization assays. Those who commit to a longer-term involvement will be encouraged to pursue an independent aspect of the project. If desired, students can also receive mentorship on post-graduate plans. Students are also expected to contribute to general lab maintenance and help with shared lab responsibilities as needed.

Our group has been developing microLED arrays for use in superresolution microscopy applications. One of the challenges, however, has been the control of a large number of access lines for the devices. In this project the student will develop a controller using a high pin count FPGA and synchronize the illumination with the camera capture of multiple images. These images will then be reconstructed into computed images with a higher resolution than possible using a widefield image. Following validation the system will be used by our collaborators to map the brain activity in freely moving mice as part of a larger program centered at Colorado State University examining the learning associated with the olfactory response of animals.

Our group engineers electronic systems to create new tools for biology and biomedicine. In nearly all cases, this includes the design of CMOS integrated circuits. Our main focus is on brain-computer interfaces, wearable imaging devices, and biosensing. Undergraduate students can become involved with all aspects of the design of the hardware and software for these systems.

This project will harness nursing knowledge in a systematic way to better capture the nuances of nursing data, leading to more comprehensive, accurate, and transparent algorithms. Additionally, the study seeks to develop scalable computational approaches to evaluate and improve the quality of data recorded by inpatient nurses and used in AI algorithms. Advanced AI methods will increasingly use data documented by nurses. Insufficient knowledge of nursing practice, nurse decision-making, and nursing workflows risks both inaccurate and undiscovered data signals. The goals of this project are to: Test and validate different computational methods (e.g., LLM, logistic regression, neural network) within a healthcare process modeling (HPM) framework applied to two AI-based use cases (classifying missing data versus missed care; classifying implicit biases) that leverage inpatient nursing and multi-modal data ready for integration with knowledge graphs. The HPM framework moves data science methods beyond transactional data analytics to model clinical knowledge, decision making, and behavior to classify and make predictions about patients that are consistent with and can enhance the quality of the data captured used to discover previously unknown patterns. Generate and validate a set of applicable knowledge graphs related to HPMs that are generalizable and valuable for the two AI-based use cases that leverage inpatient nursing and multi-modal data. Extend multi-modal approaches to HPM informed scalable computational processes combined with knowledge graphs across five additional AI-based use cases that leverage inpatient nursing and multi-modal data. Build an open-source pipeline to share and reuse the HPM informed scalable computational processes combined with knowledge graphs.

Conventional mining technologies face increasing technical challenges and expenses in processing deposits of declining grade, challenging mineralogy, or accessing ores that occur at increasingly greater depths. Economic quantities of high-value metals/critical minerals (Cu, Ni, PGM, REE, Li, Co, Au, Ag, Te, Mo, Re, W) are found within hard rock primary ores, which are igneous and metamorphic hydrothermal deposits generally found 500 to 5000 meters below surface (Haschke et al., 2016). Step changes in in situ recovery (ISR) technologies are needed to access these critical minerals deposits. The first-order major challenge facing hard rock ISR is the low porosity and low permeability of most igneous and metamorphic rocks. The proposed program seeks to test physical and chemical methods to increase in situ accessible reactive surface area and leaching of critical minerals within hard rock ores. We are seeking undergraduate student researcher(s) to assist with the following aspects of this project: Creation of rock core aggregates Testing of rock core fragmentation methods Design of proof of concept column leaching experiments Reactive transport modeling

This project aims to examine how early-stage private firms communicate with retail investors via equity crowdfunding platforms and how the availability and structure of information communicated on the crowdfunding platform influence investor participation and funding success. Equity crowdfunding platforms have revolutionized access to capital for startups, especially those facing restricted access to traditional financial institutions such as banks. By reducing information processing costs, these platforms attract a broad base of retail investors who might otherwise struggle to assess investment opportunities. However, little is known about the specific effects of different types of information provisions, such as interactive Q&A sessions between retail investors and firm founders, on investor decision-making. This study seeks to bridge that gap by providing empirical insights into how information flows impact investment behavior and overall market dynamics in equity crowdfunding.

This project focuses on the electrification of energy sectors, crucial for achieving net-zero emissions by 2050, and its challenges, particularly for marginalized communities. These communities face increased risks due to outdated grid infrastructure and volatile electricity prices, exacerbated by the integration of renewables and extreme weather events. Centered on New York City (NYC), the project aims to combine grid capacity data with demographic and building/land-use data to tackle electrification challenges. Utilizing data from Consolidated Edison, NYC's electricity provider, and extensive demographic and building datasets, the project will deliver: 1) Forecasts of electric demand increases across NYC, integrating demographic data, building characteristics, electricity consumption patterns, and heating demand models; 2) Analysis of potential network upgrade costs, identifying areas likely to be underserved by the current grid capacity; 3) Recommendations for alternative strategies like demand response or energy storage to reduce costs related to increased electrification demands.

Student researchers will apply machine learning and NLP techniques to large (>3 million) corpora of declassified documents. We will be working on a RAG system for LLMs using the documents. We are also looking for a student to assist with enhancing and maintaining a Columbia Sites website (Drupal-based CMS).

Our research is always traveling in new directions, but the following are several research areas we pursue. In all our work, we take a multi-method approach incorporating functional neuroimaging (fMRI and EEG), computational modeling, and behavioral paradigms, among other techniques.

Since 2019, faculty and students from Teacher’s College and Columbia University with an interest in neurological conditions including Parkinson’s disease and Dementia have established partnerships with Faith based organizations - St. Luke AME Church, St. John’s Baptist Church and Mt. Neboh Baptist Church, in order to deliver culturally-sensitive information and resources to educate and empower the community on how to improve their brain health. As this partnership has grown, we would like to establish a website that can host all of the resources and materials for the community, as well as serve as a platform to identify community-based programming including social programming and exercise programming that community members want to engage, in a format that is easy to navigate. The website should also include a calendar, be interactive utilizing a forum. By creating an accessible repository of resources, we will increase uptake and engagement in these resources.

Patients with Parkinsons’s Disease and their caregivers feel most medically supported when they receive personalized and culturally sensitive medical care. There is currently little existing research as to whether there are unmet needs specific to South Asian patients managing Parkinson’s Disease. This research project would aim to identify gaps in medical care provided to South Asian patients with Parkinson’s Disease. Anticipated Student Responsibilities: - Assist with background literature review - Assist with survey and overall study design - Assist with coordination of survey dissemination - Assist with organization and analysis of survey results Upon study completion, students will have the opportunity to assist with drafting a manuscript summarizing results for peer-reviewed publication

Two different heat-transport mechanisms are known in solids: in crystals, heat carriers propagate and scatter like particles, as described by the Peierls-Boltzmann equation for phonon wavepackets [Peierls, Ann. Phys., 1989]; in glasses, instead, carriers behave in a wave-like fashion, diffusing via a Zener-like tunneling between quasi-degenerate vibrational eigenstates, as described by the Allen-Feldman equation [PRL 62, 1989]. Recently, it has been shown that these two conduction mechanisms emerge as limiting cases from a unified transport equation based on the Wigner phase-space formulation of quantum mechanics [Simoncelli et al, PRX 12, 2022], which covers, on the same footing, solids ranging from crystals to glasses. Importantly, materials with an intermediate degree of disorder exist [Simoncelli et al., arXiv:2405.13161, accepted in PNAS], and these are not described by either the Peierls or the Allen-Feldman limits. This project will employ the recently developed unified theory of thermal transport and machine-learning simulation methods to deepen current understandings of the microscopic physics underlying thermal transport in hybrid crystal-glass materials with controlled degree of disorder in bond topology and geometry, which are employed as a thermal barriers in zero-carbon jet engines.