Data Scientist | 3+ Years Experience | Python, SQL, Machine Learning, Deep Learning, AWS
Actively Looking for Work
I am a data scientist at Picarro, Bangalore, working on machine learning for physical systems where the physics is well understood but the data is noisy. I did my M.Tech in Aerospace Engineering at IISc, where my thesis on robust object detection led to an IEEE paper on domain generalization. Since then I have spent about three years taking models from experiments into production.
At Picarro, I work on infrared spectral inference and atmospheric source inversion. I replaced a slow per-spectrum Bayesian fit with a single neural network forward pass, so inference now runs in real time on live sensor streams. For fenceline monitoring, I use Bayesian inference over dispersion models to localise leaks with calibrated uncertainty, and I have used diffusion models to recover time-varying emission profiles that classical optimisers could not.
Earlier, at Mercedes-Benz R&D India, I built physics-informed models for EV battery temperature prediction, a RAG tool for test engineers, and deployed models to edge devices using quantization and ONNX. I co-invented two patents there.
Outside work, I do personal projects to stay relevant as the technology keeps changing.