Prasanjit Dey

About me:

I am a Postdoctoral Researcher at Chalmers University of Technology, Gothenburg, Sweden, working on the ESA BIOMASS mission. My work focuses on algorithm development and calibration/validation of the BIOMASS Above-Ground Biomass (AGB) product: developing retrieval algorithms that estimate global forest above-ground biomass from P-band synthetic aperture radar (SAR), the observational basis for carbon stock assessment and forest-related climate mitigation monitoring.

I completed my PhD in Computer Science in June 2026 at Technological University Dublin, funded by the Research Ireland Centre for Research Training in Digitally-Enhanced Reality (D-REAL). My thesis, Monitoring and Short-term Forecasting of Atmospheric Air Pollutants Using Deep Neural Networks, was supervised by Dr. Bianca Schoen-Phelan (TU Dublin) and Dr. Soumyabrata Dev (Trinity College Dublin). It developed deep learning methods that fuse satellite retrievals, ground observation networks and ERA5 reanalysis to estimate and forecast near-surface pollutants such as PM2.5, NO₂ and O₃.

I build machine learning systems for environmental prediction, and I care as much about how far a forecast can be trusted as about how accurate it is. My research centres on three strands:

- Physics-informed spatiotemporal modelling: hybrid architectures that couple neural networks with explicit diffusion–advection dynamics (e.g. NeuroDDAF, AirGRU), so predictions stay physically consistent rather than merely well fitted. - Forecasting in data-scarce settings: parameter-efficient adaptation of pretrained large language models (LoRA/rsLoRA) for few-shot and zero-shot forecasting (e.g. GPT4AP, One-for-All). - Uncertainty quantification: evidential and conformal calibration, so that every forecast carries an honest interval a decision-maker can act on.

Highlights:

- 25 peer-reviewed outputs: 14 journal articles (10 published, 4 under review) and 11 refereed conference papers, in venues including IEEE JSTARS, IEEE GRSL, Process Safety and Environmental Protection, and IGARSS. - 2 granted Indian patents: Digital Mine Using Internet of Things (2026) and Location Tracking System for Indoor Environment (2023). - Open-source code with pinned dependencies, seeded runs and container images: NeuroDDAF and One-for-All. - CSIR-CIMFR Dr. Adinath Lahiri Award for outstanding research publication (2021).

News:

- Sep 2026: Joined Chalmers University of Technology as a Postdoctoral Researcher on the ESA BIOMASS mission. - Jun 2026: Successfully defended my PhD thesis at TU Dublin. - Feb 2026: Indian patent Digital Mine Using Internet of Things granted. - Aug 2025: Presented two papers at IEEE IGARSS 2025, Brisbane, Australia. - May 2025: Started a visiting research placement at TU Delft, Geoscience and Remote Sensing.

Experience:

1. Postdoctoral Researcher at Chalmers University of Technology, Sweden from 2026/09–Present. ESA BIOMASS mission: developing retrieval algorithms for global forest above-ground biomass from P-band SAR; building processing and inference pipelines over multi-temporal, multi-polarisation SAR stacks; and calibrating and validating the operational product against in situ and airborne reference data, with quantified uncertainty.

2. Doctoral Researcher at Technological University Dublin, Ireland from 2022/09–2026/06. I developed physics-informed, transformer, graph and generative architectures for multistep spatiotemporal forecasting of air pollutants, as well as parameter-efficient adaptation of pretrained LLMs for data-scarce regions. I also mentored junior doctoral researchers.

3. Visiting Research Scientist at TU Delft, Netherlands from 2025/05–2025/08. Geoscience and Remote Sensing Department: I led development of a coupled dynamics-and-learning framework for satellite-based air quality forecasting, together with Dr. Angela Meyer.

4. Teaching Assistant at Technological University Dublin, Ireland from 2023/01–2025/05. I supported labs, tutorials and assessments in Operating Systems, Python, Secure Programming, Robotics and Software Engineering.

5. Junior Research Fellow at National Institute of Technology Jamshedpur, India from 2020/12–2022/08. I applied deep learning to environmental prediction from satellite and radar observations, and built curation, quality-control and pipeline tooling for large geospatial and meteorological datasets.

6. Research Engineer (Project Assistant Level-III) at CSIR-Central Institute of Mining and Fuel Research (CSIR-CIMFR), India from 2018/12–2020/10. I was technical lead for deployed environmental monitoring and hazard-prediction systems (IoT sensor networks with real-time inference), delivered findings to a national ministry, safety regulators and site operators, and am co-inventor on two granted patents.

7. Assistant Professor at Ramgovind Institute of Technology, India from 2017/02–2018/07. I was sole lecturer for four undergraduate courses (computer networks, operating systems, database management, algorithms and data structures) and supervised six final-year projects.

Education:

1. PhD in Computer Science at Technological University Dublin, Ireland, from September 2022 – June 2026. Thesis: Monitoring and Short-term Forecasting of Atmospheric Air Pollutants Using Deep Neural Networks.

2. MTech in Information Technology from Maulana Abul Kalam Azad University of Technology, West Bengal, from June 2014 – September 2016.

3. BTech in Computer Science and Engineering from West Bengal University of Technology, Kolkata, India, from June 2010 – May 2014.

Research Interests:

Earth Observation & SAR, Climate AI, Physics-Informed Machine Learning, Spatiotemporal Forecasting, Foundation Models for Time Series, Uncertainty Quantification

Programming Languages:

Proficient with: Python, C, LaTeX, SQL, Bash

Familiar with: MATLAB, R, Java, C++

Frameworks & Libraries:

PyTorch, TensorFlow, NumPy, SciPy, pandas, xarray, scikit-learn

AI/ML Techniques:

Physics-Informed ML, Neural ODEs, Graph Neural Networks, Vision Transformers, Generative Models, LLMs & Parameter-Efficient Fine-Tuning (LoRA), Uncertainty Quantification, Time-Series Forecasting, Computer Vision

Data & Tools:

ERA5, Sentinel, MODIS, P-band SAR; NetCDF, HDF5, GeoTIFF, Zarr; Linux, HPC, CUDA (NVIDIA certified), Docker, Git, OpenCV

CV

Find attached the PDF version of my CV:
English version: CV

Update: 2026/09/23