Relevant Coursework

Introduction to Deep Learning for Engineers - CMU

Intermediate Deep Learning - CMU

Machine Learning and Artificial Intelligence for Engineers - CMU

Probability and Estimation Methods for Engineering Systems - CMU

Statistical Learning and Modeling - CMU

Research Interests

• Applying machine learning methods to physics-driven climate models.

• Building machine/deep learning-based air pollution forecast models with novel training features

• Developing low-cost methods to measure atmospheric pollutants in the Global South

• Source apportionment of emission sources using pollutant measurements from low-cost and reference monitors

Research Experience

Machine/deep learning-based PM2.5 forecast models with GEOS-CF and MODIS AOD features

I am developing a deep learning-based PM2.5 forecast model for Allegheny County (Pennsylvania, USA) using PM2.5 measurements from a low-cost sensor network of 50 Real-time, Affordable, Multi-Pollutant (RAMP) monitors as ground truth. Each RAMP measures CO2, CO, NO2, and PM2.5 every minute. I am building a deep Transformer Model using downscaled 3 km × 3 km (originally, 25 km × 25 km) gridded 5-day forecast for air quality (AQC) and meteorological (MET) parameters from GEOS-CF model, and 3 km × 3 km gridded AOD from the MODIS satellite instrument as training features. Furthermore, I compare its performance with conventionally used statistical model: Autoregressive Integrated Moving Average (ARIMA) and  deep learning model: Long Short-Term Memory (LSTM). This study aims to forecast instances of elevated pollution in accordance with the Mon Valley Episode Rule and mandate preemptive emission reductions for industries and facilities in the Monongahela Valley to avoid significant pollution levels during these episodes.

Developing low-cost methods to measure atmospheric pollutants in the Global South

During my Ph.D., I worked with Prof. Albert Presto on devising a cost-effective computer vision-based technique (Journal article) that exploits existing Beta Attenuation Monitors (BAMs) at US Embassies to estimate atmospheric black carbon (BC) concentrations. The technique takes photos of hourly particle deposits on used BAM tapes and does image processing of the photos to estimate BC levels, a tracer for combustion emission sources. Through this work, I acquired hands-on experience in implementing image-processing algorithms for atmospheric science research.

I have submitted another journal article (available at ChemRxiv) to Environmental Science & Technology, which is currently under review. This paper validates our image reflectance-based method with a reference BC monitor at an African site (Addis Ababa, Ethiopia), and presents continuous hourly measurements for ambient BC from cities in Sub-Saharan Africa using our method. These cities include Abidjan (Côte d'Ivoire), Accra (Ghana), and Addis Ababa. Additionally, I used our ground-level BC measurements to evaluate the performance of a widely used CTM, Goddard Earth Observing System Composition Forecast (GEOS-CF) at a 25 km × 25 km grid resolution. We found that GEOS-CF significantly underestimated ground level BC (↓77% in Accra) indicating the importance of ground-level monitors for calibrating these models in the African continent.

ASCENT project

I have been involved in other research projects during my PhD program. Majorly, I participated in establishing and analyzing data for the Pittsburgh site, a part of the Atmospheric Science and Chemistry mEasurement NeTwork (ASCENT) project. ASCENT is a nationwide collaborative effort across 12 sites in the U.S. with measurements of fine aerosol chemical composition and properties with advanced instruments: ACSM (organics and non-refractory inorganics), XAct (metal speciation), AE33 (BC), and SMPS (particle size distribution). 

VOC measurement with Mobile van

I was also a part of a mobile air pollution monitoring campaign, led by another Ph.D. student in Presto group, aimed at measuring volatile organic compounds (VOCs) from cooking and asphalt emissions in Pittsburgh (\textit{manuscript under preparation}). My duties mainly included working with a team to setup the instruments in a van, collecting VOC data with the van near restaurants and asphalt plants, and analyzing instrument data.

Education

• PhD Student in Mechanical Engineering, Carnegie Mellon University (2020 - Present) 

• M.Phil. in Environment Science, Policy, and Management, Hong Kong University of Science and Technology (2018 - 2020)                 

• M.Sc. in Environmental Engineering and Management, Hong Kong University of Science and Technology (2017)

• B.Tech. in Civil Engineering, Indian Institute of Technology Delhi (2011 - 2015)                                                                      

Awards and Honors

• Sep 2023: Selected for the AccelNet Early Career Researcher (ECR) Professional Development Workshop, Arlington, Virginia

• 2022 - 2023: Dowd Fellowship, CMU

• 2022: Milton Shaw PhD Research Award, Department of Mechanical Engineering, CMU

• 2020: Selected by HKUST to attend for the Global Young Scientists Summit (GYSS), National Research Foundation, Prime Minister’s Office, Singapore

• 2018 - 2020: Postgraduate Studentship (PGS)

• Fall 2019: University Grants Committee (UGC) Research Travel Grant

• Fall 2019: Division of Environment and Sustainability (ENVR) Research Travel Grant

• Feb 2018 - May 2018: Innovation and Technology Fund (ITF) Project Internship Awardee, Hong Kong

• 2016-2017: M.Sc. Excellent Student Scholarship, School of Engineering, HKUST

• 2016-2017: Entrance Scholarship, School of Engineering, HKUST

• 2011-2015: Ministry of Human Resource Development (MHRD) Scholarship, Indian Institute of Technology Delhi, India 

Teaching

• Fall 2021 - Present: Future Faculty Career Program at CMU, designed to help graduate students develop and document their teaching skills in preparation for a faculty career

• Aug 2021 - May 2022: Undergraduate Peer Tutor at CMU 

  • • • • • Physics I for Science Students (33121)

          • Physics II for Biological Sciences and Chemistry Students (33122)

          • Physics I for Engineering Students (33141)

          • Physics II for Engineering and Physics Students (33142)

          • Calculus (21111-122)

          • Differential Equations (21260)

• Spring 2023 - Teaching Assistant, Renewable Energy Engineering - 24792, CMU

• Spring 2022 -  Teaching Assistant, Fluid Mechanics - 24231, CMU 

• Fall 2019 - Teaching Assistant, GIS for Environmental Professionals - EVSM 5240, HKUST

• Spring 2019 - Teaching Assistant, Carbon Emission Trading - ENVR 6090A, HKUST