• Implemented data science and MLOps processes for data exploration, model development, deployment, inference, monitoring, and retraining. • Conducted research to discover use cases in energy optimization and predictive maintenance leveraging ML. • Assisted other data science teams in transitioning their models from development to production. • Developed models focused on predicting and optimizing energy usage. • Facilitated the deployment process of data models from development stages to production environments. • Established frameworks for continuous monitoring and periodic retraining of deployed models. • Collaborated with various departments to ensure data science initiatives aligned with organizational goals. • Created predictive maintenance models to enhance equipment efficiency and minimize downtime.

• Lead the development of earning forecast models for several publicly-traded companies based on a proprietary • Achieved superior accuracy compared to publicly available analyst estimates • Closely collaborated with the business team and domain experts to perform meaningful data analysis and feature engineering for the problem of earnings forecasting • Developed automatized retraining and inference pipelines for future fiscal quarters for ease of use

• Lead the development of several NLP models that extract human trafficking and modern human slavery-relevant information from news data sources • Developed several NLP models for text classification and tagging • Closely collaborated with the business team, product owners, other tech teams, and data labelers to ensure the alignment between AI technology and other business operations. • Fine-tuned an LLM for MHSHT domain • Developed robust training and inference pipelines, provided documentation so that the organization can continue to improve the AI models with minimal technical knowledge • Integrated the AI models into the organization’s Traffick Analysis Hub to enable its users to leverage the power of AI to monitor MHSHT activities in real time. Tech stack: Python, pytorch, huggingface, OpenAI-API, transformers, IBM Cloud, Docker, Kubernetes, MongoDB, NLP, NLU, LLMs

• Lead the implementation of a Bayesian Spatiotemporal-Aware Climate Model Ensembling Method for a specialized hardware platform in C • Supported the compiler team in the development of specialized routines for Bayesian inference by performing hardware-specific analysis of neural networks • Contributed to DSP and ML library for a specialized CPU with the implementation of complex algorithms like Radix-2 FFT, Levenberg-Marquardt algorithm, non-linear least squares optimization, IIR filter design procedure, and filtering, and others. • Improved an existing signal processing pipeline for underwater cable displacement in MATLAB • Ported the said algorithm from MATLAB to C for a specialized hardware platform and optimized its performance