Senior researcher in Mediatek Research, Mediatek's AI research arm: worked on algorithmic improvements for generative models like diffusion models and LLMs, as well as applications such as training foundation models for modems and building autonomous agents for performing tasks on a phone.

I applied different machine learning based thermoacoustic instability prognosis tools to pressure signal data from an experimental rig. I am also working on synchronizing experimental time series using autoencoders.

Demonstrator for 3F8 Statistical Inference course (Lent 2019), 1A Dimensional Analysis Fluids Lab (Michaelmas 2020, 2021) and 1B Boundary Layers Lab (Michaelmas 2020).

Ph.D. in engineering under the guidance of Professors Carl Rasmussen and Matthew Juniper. My thesis focused on applying Bayesian machine learning tools to model, prognose and design against thermoacoustic instabilities. Was also involved in collaborative projects to optimally combine chemistry-climate model predictions with Bayesian Neural Networks and to apply Physics-Informed Neural Networks to denoise magnetic resonance velocimetry data.

We used multimodal sensor data from BKD, an experimental rocket engine combustor operated by the German Aerospace Center, to forecast thermoacoustic instabilities using an autoregressive Bayesian neural network model.

Developed ML methods to predict electronic density in molecules, explored novel approaches using optimal transport theory, and collaborated on calculating neuronal cell membrane capacitance via Molecular Dynamics simulations.

Worked on constructing automated Markov state models for molecular dynamics simulations, with a focus on protein aggregation and enzyme catalytic activity. Published work in well-regarded journals and presented at international conferences.

Bachelor's thesis on micro-combustion under Prof. S.K. Som; analyzed flame stability & efficiency of micro-combustors using ANSYS Fluent, revealing insights into inlet swirl numbers' impact.

DAAD-funded research internship exploring synthetic cartilage for arthritis treatment: developed novel cellulose fiber-reinforced porous PDMS material and worked on experimentally analyzing its mechanical and tribological properties.

Collaborated with Prof. Arup Kumar Das on fire modeling using FDS, specializing in compartment fires and pre-flashover forensic patterns; presented findings at a national conference.

Worked with the Microfluidics Research Group under the supervision of Suman Chakraborty, a professor in the Department of Mechanical Engineering. My project involved Molecular Dynamics simulations of binary fluid mixtures performed using the GROMACS software package. I simulated the flow of such a mixture through a channel with patterned wettability and studied how that pattern enhanced the mixing.

Identifying bottlenecks in the 5 star candy production process, performing root cause analysis of production issues.

Researched light transmission systems like optical fibers and light rods for designing an efficient photo-bioreactor that would optimize the growth of fuel-producing algae.