Projects

Every project here started with a problem worth solving!

High-stakes systems don’t forgive bad engineering. Every project here was built under that constraint — where uncertainty is costly, the data is messy, and the margin for error is real. From autonomous agent pipelines to climate prediction systems, the work spans domains but shares a single standard: it has to work.

 

Browse the full portfolio below, or open any project for the complete case study.

Tier 1

Agentic AI Systems

Autonomous pipelines · Multi-agent coordination · Real-world task execution

View All Projects →
Agentic AI Systems Research & Knowledge Work
🔍

Deep Researcher

OpenAI SDK · Fully Agentic

A fully autonomous multi-agent research pipeline where agents control the entire search strategy — topic decomposition, query formulation, source evaluation, and synthesis — without user intervention. Delivers a structured report by email upon completion.

⭐ Fully autonomous — agents determine search strategy without inputs

OpenAI SDK Multi-Agent Gradio
View Project →
Agentic AI Systems Travel & Logistics
✈️

Trip Planner

CrewAI Hierarchical

A hierarchical multi-agent trip planning system where a manager agent coordinates specialized sub-agents — destination researcher, activity planner, budget optimizer, and itinerary writer — to produce a complete, budget-aware travel plan with PDF output.

⭐ Hierarchical agent coordination — manager directing 4 agents

CrewAI Multi-Agent Gradio
View Project →
Agentic AI Systems Personal AI & Career Tech
🤖

Career Digital Twin

OpenAI SDK · RAG

An AI-powered career assistant that ingests a structured resume and career data to answer queries about professional background, research, publications, and experience — deployed as a conversational Gradio app on Hugging Face Spaces.

⭐ Turns a static CV into a live, queryable professional presence

OpenAI SDK RAG ChromaDB Gradio
View Project →

Tier 2

Applied ML Systems

High-stakes domains · Production-grade · Research-validated

Applied ML Systems Climate Resilience & Infrastructure
🌊

Calgary Flood Prediction System

Deep Learning · Production ML System

A fully productionized deep learning pipeline that predicts river flood inundation depth across multiple locations within Calgary. Ingests meteorological and hydrological data and outputs spatial flood depth forecasts — giving emergency planners actionable intelligence before water levels rise.

⭐ 89% production accuracy · Fully productionized and operational for the City of Calgary

TensorFlow TFX GCP Vertex AI Keras Python
View Project →
Applied ML Systems Climate Resilience & Infrastructure
🔥

Alberta Wildfire Prediction System

Deep Learning · Production ML System

A province-scale deep learning pipeline predicting wildfire occurrence and spread risk across Alberta using meteorological data, vegetation indices, and topographic features. Outputs geospatial fire risk maps to support resource pre-positioning and public alert decisions.

⭐ Province-scale prediction covering the entire Province of Alberta with geospatial risk maps

TensorFlow TFX GCP Vertex AI Keras Python
View Project →
Applied ML Systems Logistics & Transportation
🚁

Drone Risk Prediction & Mitigation System

ANNs · Closed-Loop Decision System

A deep learning-based risk assessment system for autonomous drone delivery of hazardous materials in urban environments. Trains on operational data to produce dynamic safety assessments informing route planning, regulatory compliance, and incident prevention.

⭐ Real-time deep learning inference on 12 operational features — from flight altitude to hazmat type

TensorFlow Keras ANNs Python
View Project →
Applied ML Systems Infrastructure Risk & Project Controls
🕸️

Infrastructure Risk Intelligence System

Graph-Based ML · Multi-Objective Optimization

A graph-based ML and deep learning pipeline that predicts cascading infrastructure project risks using Genetic Algorithm tuning and NSGA-II optimization. It captures how interdependent risks propagate across complex project networks to support better project controls.

⭐ Up to 97% prediction accuracy · Models systemic risk cascades across interdependent project networks

scikit-learn Graph Algorithms Genetic Algorithm
View Project →

Tier 3

MLOps & Production Pipelines

End-to-end deployment · Orchestration · Monitoring & automation

MLOps & Production Pipelines ML Pipeline Orchestration
⚡

ChurnFlow: End-to-End MLOps Pipeline

MLOps · Full Lifecycle Pipeline

A production-ready MLOps pipeline for telecom customer churn prediction. Covers the full lifecycle — model training, experiment tracking, automated CI/CD, real-time data drift detection with Evidently AI, and FastAPI serving — built to the standards of a production engineering team.

⭐ Full MLOps: from training to monitored production deployment with drift detection

FastAPI EvidentlyAI MLflow Docker GitHub Actions
View Project →
MLOps & Production Pipelines Retrieval-Augmented Generation
📄

RAG App: Hybrid Retrieval Question Answering

RAG · Hybrid Retrieval

A production-style RAG system that ingests PDF documents, indexes them using hybrid BM25 lexical and FAISS vector search, and serves precise, source-grounded answers via a FastAPI endpoint — built with enterprise document intelligence architecture patterns.

⭐ Hybrid BM25 + FAISS retrieval for high-accuracy, source-grounded document QA

FastAPI FAISS BM25/Whoosh LLMs PyPDF Python
View Project →
MLOps & Production Pipelines Airflow-based Pipeline Orchestration
🔄

AirflowTFX: End-to-End MLOps Pipeline

TFX · Apache Airflow

A TFX-based ML pipeline orchestrated with Apache Airflow — automating data ingestion, schema validation, feature transformation, model training, and evaluation through modular DAG-based tasks. Runs in Dockerized environments for clean portability across teams.

⭐ Modular, reproducible ML workflow with baked-in data validation, transformation, and evaluation using Airflow

TFX Apache Airflow TFDV TFT TFMA
View Project →
MLOps & Production Pipelines Kubernetes-native Pipeline Orchestration
☁️

KubeTFX: Kubernetes-Native ML Pipeline

TFX · Kubeflow · Kubernetes

A TFX pipeline deployed on Kubeflow Pipelines running on a local Minikube Kubernetes cluster. Components are containerized with Docker and deployed using YAML-based configuration — simulating production-ready cloud-native ML workflows with elastic scaling.

⭐ Cloud-native pipeline orchestration with Kubernetes-native scaling and containerized components

TFX Kubeflow Minikube Docker TFDV
View Project →

Tier 4

Open-Source Tools

Publicly released · Community adopted · Built to last

Open-Source Tools ML Tooling
🔧

Hyperparameter Optimizer

Python Package · PyPI

A Python library for automated hyperparameter tuning using Particle Swarm Optimization and Pattern Search. Acts as a drop-in replacement for GridSearchCV, converging faster to optimal configurations with native scikit-learn compatibility.

⭐ 2,000+ downloads by ML practitioners worldwide. Faster convergence than grid search

Python scikit-learn PSO Pattern Search PyPI
View Project →
Open-Source Tools ML Tooling
📊

TFX Evaluator

Python Package · PyPI

A custom TFX component that fills a gap in the standard ecosystem — computing per-output and global metrics for multi-output TensorFlow models and writing TFMA-compatible JSON artifacts for downstream pipeline integration.

⭐ Fills a critical gap in the TFX ecosystem · Adopted for production pipeline reproducibility

Python TensorFlow TFX TFMA Pipelines PyPI
View Project →
Open-Source Tools Geospatial ML
🗺️

GRAX

Python Package · PyPI

A lightweight Python library that transforms geospatial shapefiles into ML-ready graph networks using NetworkX. Bridges the gap between GIS data formats and graph-based ML workflows by parsing shapefile geometry into structured representations.

⭐ Bridges GIS and graph ML — no standard tooling existed for this workflow before

Python NetworkX Fiona Shapely NumPy PyPI
View Project →

Tier 5

ML Engineering

Focused explorations · Algorithm benchmarking · Applied domains

View All Projects →
ML Engineering Computer Vision
🎥

DeepVision: Video Analysis

ConvLSTM Architecture

A ConvLSTM deep learning model that combines spatial feature extraction with temporal sequence modeling for automated video classification. Processes raw video frames through an OpenCV pipeline and outputs class predictions for video-heavy operational workflows.

⭐ 84% accuracy, 97% AUC. Combines spatial and temporal learning

TensorFlow OpenCV ConvLSTM CNN Python
View Project →
ML Engineering Healthcare
🏥

Coronary Heart Disease Risk Classifier

Ensemble Classification

A binary classification system predicting 10-year coronary heart disease risk from patient vitals and lab results. Benchmarks Random Forests against Mixture Discriminant Analysis across 10 random seeds to ensure consistent, reliable predictions for clinical decision support.

⭐ 84.3% classification accuracy. RF outperformed MDA with lower variance

R Decision Trees Random Forest MDA
View Project →
ML Engineering Structural Engineering
🏗️

Concrete Compressive Strength Predictor

Tree Ensemble Regression

A tree ensemble regression pipeline predicting concrete compressive strength from mix ingredients — replacing time-consuming lab testing with rapid ML predictions. Benchmarks Bagging, Random Forests, and Gradient Boosting on 1,030 specimens across 8 predictors.

⭐ Boosting significantly outperformed Random Forest and Bagging

R Gradient Boosting Random Forest Bagging
View Project →