- Researching approaches to evaluate Generative AI systems for trustworthiness (performance, bias, safety, robustness, etc).- Collaborating with various clients to assess their machine learning systems for potential performance and ethical issues according to the NIST Risk Management Framework (among others).

Alexa AI teamWorked on mitigating hallucinations (via fine-tuning and novel prompt approaches) in LLMs for knowledge-grounded (NL2SQL) dialogue systems.

As part of data engineering team, I worked on developing ETL and analysis pipelines using Airflow for terabytes of streaming data.

◦ Manga Colorization: Worked on developing a deep learning model to color black & white Manga (japanese comics) images. Later, developed ML workflow for deployment.◦ Backend development: Worked on a Django-based web application where I worked on developing core modules, designed and implemented the backend architecture for asynchronous tasks (using celery) and implemented the CI/CD pipelines to automate the code deployment process.

◦ Standard REST API Definitions: Worked on a project to convert the Swagger API Definition Standard to RAML API Definition Standard and vice versa. The code is shared on my Github repository.◦ RPC Frameworks: Added RPC support for the corporate’s software using gRPC and Apache Thrift Frameworks.