Implemented, trained, and tested an optimized 40M parameter document segmentation model(DocStructNet), with 97% accuracy on the DocLayNet and PubLayNet datasets

Used the AWS fmeval library, along with ECS, EC2, S3 and SageMaker to deploy a government chatbot with an evaluation pipeline

Reworked a Transport Canada Protected B application to replace AzureBlobService with a local version, and produced an exhaustive report on its functionality and security

Completed tickets in an Azure DevOps environment, interfacing with Azure GPU VMs, Azure Container Registry, and Azure SQL Database

Finetuned an open source LLM from the Hugging Face model library that enabled GPU costs to be reduced by 90%

Cleaned CIS Controls data from SCF, SOC II, CMMC and other cybersecurity frameworks

Used LangChain and HuggingFace embeddings to build a file-based chatbot that was able to effectively interact with clients and respond to queries

Studied Figma best practices and built several mock-ups to present to the client

Translated Figma components into functional React Native components

Used a provided REST API to integrate live data into the app