Deployed large language models (LLMs) using HuggingFace's Text Generation Inference (TGI) and NVIDIA's Triton Inference Server with TensorRT-LLM on an AWS EC2 machine. Conducted performance comparisons between these two inference servers to evaluate their efficiency and scalability. Additionally, documented the deployment processes for both TGI and TensorRT-LLM to facilitate replication. The documentation included a detailed analysis of HuggingFace's TGI, as well as the parameters that can enhance performance when using TensorRT-LLM.

Composed and moderated data collection methods for events delivered by the club.

Developed an Android application for automatic hotspot creation, in order to calibrate and improve the employee locator system that was in progress in the office.