Skip to content
View tuilachit's full-sized avatar
0 followers · 1 following
  • Sydney
  • 19:04 (UTC +10:00)

Block or report tuilachit

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
tuilachit/README.md

Luke Nguyen

AI and full-stack developer based in Sydney, Australia. I build practical agentic systems, machine-learning products, and data-driven applications with an emphasis on safety, evaluation, and useful outputs.

Featured projects

Project Impact and engineering focus Stack
NSW Vegetation Classifier End-to-end geospatial inference application using a two-stage vegetation-segmentation pipeline, FastAPI backend, and inspectable raster outputs. Python, TensorFlow, FastAPI, GeoTIFF
SignalDesk AI communications agent with read-only Gmail and Slack connectors, explicit approval gates, dry-run execution, and persistent audit logs. TypeScript, React, OpenAI, SQLite
PPP Candidate Research Agent Converts candidate CSVs into source-grounded research briefings with structured JSON output and uncertainty handling. Python, Claude, Streamlit
Career Compass Full-stack career-guidance platform with a Wagtail/Django backend and Next.js frontend. Python, Django, Wagtail, Next.js
Toxic Comment NLP Compares BiLSTM and DistilBERT multi-label classifiers; the selected model reached 0.789 micro-F1 and 0.694 macro-F1. PyTorch, Transformers, Gradio
Reasi AI Responsive product and waitlist experience for an AI grocery-planning agent. Live demo. React, Vite, Supabase, Vercel

What I bring

  • Agent workflows with allowlists, verification gates, human approval, and audit trails
  • Applied ML and NLP with reproducible experiments and task-appropriate evaluation
  • Full-stack delivery across React, Next.js, TypeScript, Python, Django, and APIs
  • Data products that turn ambiguous source material into dashboards, reports, and structured artifacts

Technical toolkit

  • Languages: Python, TypeScript, JavaScript, SQL, Dart
  • AI/ML: PyTorch, TensorFlow, Transformers, OpenAI API, Claude API, evaluation workflows
  • Frontend: React, Next.js, Vite, Tailwind CSS, Framer Motion
  • Backend and data: FastAPI, Django, Wagtail, GraphQL, PostgreSQL, SQLite, Supabase
  • Data and geospatial: Pandas, Tableau, Excel, Rasterio, GeoPandas
  • Workflow: Git, GitHub Actions, Vercel, Streamlit

Currently building

I am focused on interview-ready AI products that demonstrate engineering judgment: clear problem framing, measurable results, safe tool use, reproducible setup, and honest discussion of limitations.

If that overlaps with what your team is building, feel free to reach out through GitHub.

Pinned Loading

  1. ToxicComment_NLP_Project ToxicComment_NLP_Project Public

    PyTorch NLP project comparing BiLSTM and DistilBERT for multi-label toxic comment classification.

    Jupyter Notebook 1

  2. Careercompass Careercompass Public

    Full-stack career guidance platform using Wagtail/Django APIs and a Next.js 15 frontend.

    Python

  3. hourglassai hourglassai Public

    SignalDesk: TypeScript AI comms agent with Gmail/Slack connectors, audit logs, safety gates, and a React dashboard.

    TypeScript

  4. NSW_Vegetation_Classifier NSW_Vegetation_Classifier Public

    Deep learning system for classifying NSW vegetation from multi-channel geospatial imagery with TensorFlow and FastAPI.

    Python

  5. ppp-candidate-research-agent ppp-candidate-research-agent Public

    Claude-powered Python research agent that turns candidate CSVs into evidence-grounded funds-management briefings.

    Python

  6. reasi_ai reasi_ai Public

    React/Supabase waitlist site for Reasi, an AI grocery agent that plans meals, builds lists, suggests swaps, and prepares checkout.

    CSS