I enjoy building AI-driven applications

Projects

Apple Notes MCP

A Model Context Protocol (MCP) server that provides CRUD tools for Apple Notes using AppleScript integration.

Python

FastMCP

AppleScript

SecondPlace

Built an AI-driven marketplace for secondhand clothing, automating listing with Snap-Check-List.

Python

YOLO

CVAT

vLLM

FastAPI

Flutter

AWS

Docker

RAG Workflow

A RAG (Retrieval-Augmented Generation) workflow using LangGraph for orchestration and Qdrant for vector storage.

Python

LangGraph

Qdrant

FastAPI

Next.js

Mistral Reasoning Finetune

Finetuned a Mistral-7B-Instruct model using PEFT (LoRA) and Unsloth for chain-of-thought (CoT) and reasoning tasks via SFT.

Python

PyTorch

HuggingFace

Transformers

DBS Mobile App

Developed a mobile app for DBS in collaboration with Google, that utilizes various Google Vertex AI and external APIs, to expedite user onboarding.

Ruby on Rails

React Native

Google Cloud Platform

Vertex AI

Experiences

Handshakes by DC Frontiers

July 2024 - September 2024

Conducted comprehensive research to optimize inference speeds and identify the most efficient ways to deploy Large Language Models (LLMs) by evaluating various inference libraries and quantized model formats
Established a scalable experimentation framework that allows for rapid experimentation with emerging LLM models and inference libraries, enhancing performance on tasks such as summarization and classification

Handshakes by DC Frontiers

Feb 2024 - June 2024

Spearheaded a project to develop a web crawler, automating data collection from 12 websites, encompassing requirements gathering, system architecture design, implementation, rigorous testing, and deployment
Built the API Developer Portal for Handshakes using Docusaurus and Keycloak for identity and access management, ensuring secure routes on the website