Ali Zamani | Senior ML/AI Engineer

About Me

Production AI at scale

M.Sc. in Computer Science (University of Alberta) · Senior ML/AI Engineer at Loblaw Digital, Toronto.

I build production AI systems end to end — from planner-based agentic architectures and hybrid-search RAG to standardized ML pipelines and GenAI frameworks on GCP.

At Loblaw Digital, I lead enterprise AI initiatives — shared agent infrastructure, production retrieval systems, and next-generation agentic platform architecture for retail at scale.

Previously at Priceline (Booking Holdings), I built Vertex AI ML pipelines, a scalable GenAI framework with LLM-as-a-Judge validation, NLP features including hotel review summarization that drove a 30% conversion uplift in A/B testing, and a flight price prediction system using time series forecasting on Vertex AI and BigQuery.

Platform Engineering

Agent Tools Package with MCP support — centralized agent tooling adopted across AI product teams.

Retrieval & RAG

Production hybrid-search system — 97% hit rate with projected $75K–$85K annual cost savings vs. managed alternatives.

Cross-Functional Delivery

Reusable search and platform services built with partner teams across the organization.

Technical Leadership

Led agentic platform architecture and guided engineering teams through a major transformation.

Career

Professional Experience

From agentic AI and production RAG to GenAI frameworks and ML pipelines on Vertex AI.

Loblaw Digital

Senior ML/AI Engineer

Jun 2025 – Present · Toronto, ON

Proposed and led agentic platform architecture with planner-based agents, multi-layer guardrails, and long-, short-, and episodic memory — decomposing work into parallel streams and providing technical direction for the broader team.
Proactively identified overlapping agent tooling across AI teams and designed a centralized Agent Tools Package from scratch — architecture, standards, release process, onboarding, and MCP (Model Context Protocol) integration. In production with flagship AI agents.
Led build-vs-buy evaluation and delivered a production hybrid-search RAG system with custom retrieval architecture, indexing pipelines, and APIs — achieving 97% hit rate, 78% precision, and 97% recall with projected $75K–$85K annual cost savings over managed alternatives.
Partnered with the search engineering team to build a cross-domain search routing service — a reusable platform component supporting multiple search experiences.

Priceline (Booking Holdings)

Machine Learning Engineer

Apr 2023 – Jun 2025 · Toronto, ON

Built a standardized ML pipeline for preprocessing, training, and online/offline deployment on Vertex AI, featuring modular design, feature store integration, versioning, and backfilling. Orchestrated the pipeline with a reusable Airflow DAG.
Designed a scalable GenAI framework on GCP with CI/CD integration, featuring a robust Inference module with rate limiting, monitoring, parallel processing, and use-case-specific configuration, and a Validation module using LLM-as-a-Judge for automated evaluation and continuous quality feedback. Leveraged LangChain for orchestrating dynamic LLM workflows and integrating retrieval-augmented generation capabilities.
Delivered a hotel review summarization use case combining sentiment analysis and summarization, with an algorithm to surface high-impact reviews. Achieved a 30% conversion rate uplift through A/B testing.
Built a flight price prediction use case using time series forecasting. Deployed on Vertex AI with data from BigQuery, enabling data-driven pricing and integration with product features.
Supported a customer-facing chatbot using Vector RAG, fine-tuned models, and AI agents to enhance customer interactions through contextual understanding and dynamic response generation.

AltaML

Data Scientist (Contract)

Sep 2022 – Dec 2022 · Calgary, AB

Built a complete ML pipeline on Azure for rock permeability prediction using LightGBM and XGBoost, achieving 94% accuracy.
Implemented LightGBM and XGBoost models for predicting permeability of rock core images with 94% accuracy, saving upwards of $10 million for the client.
Developed the ML pipeline from scratch on Azure and conducted error analysis to further improve model performance and robustness.

MIRA Chatbot

Data Scientist & Chatbot Developer

Jan 2021 – Sep 2022 · Edmonton, AB

Dept. of Computing Science, University of Alberta & Amii

Developed and deployed the back-end and front-end of the MIRA chatbot (mymira.ca), enhancing user experience with advanced NLP.
Evaluated various Recurrent Neural Network language models for intent detection and entity extraction, achieving an F1-score of 97% for intent detection and 83% for entity extraction.
Applied data augmentation techniques to improve training data quality and model performance.
Applied sentiment analysis methods to assess user sentiments and tailored chatbot responses accordingly, improving user experience.
Trained and managed two undergraduate students on the MIRA chatbot team.

Portfolio

Selected Projects

Production systems from current role, hackathon wins, and open-source ML tooling.

Enterprise AI · Current Role

MCP Agentic AI Platform

Agent Tools Package

Proactively identified overlapping agent tooling across AI teams and designed a centralized package from scratch — architecture, standards, release process, onboarding, and MCP integration. In production with flagship AI agents.

RAG Retrieval

Production RAG System

Hybrid-search retrieval platform with 97% hit rate and projected $75K–$85K annual cost savings — custom-built for quality, flexibility, and long-term scalability.

Architecture Guardrails

Agentic Platform Architecture

Next-generation agentic system with planner-based agents, guardrails, and multi-tier memory architecture.

Other Projects

Google Hackathon Vector Embeddings

Neighborhood Recommendation System

Leveraged vector embeddings to suggest neighborhoods to users based on their preferences, enhancing personalized travel experiences.

GPT-4 Google Hackathon

Travel Assistance Chatbot

Developed a chatbot using GPT-4 to assist customers in finding their desired travel destinations, significantly improving user engagement and satisfaction.

Python Open Source

ML Pipeline Template

Developed an ML pipeline template to create a user-friendly utility that drastically speeds up the development and implementation of machine learning models for all sorts of various problems.

Source Code

Azure BERT Microsoft Hackathon

Burnout Detection System

Built an ML pipeline on Azure to detect burnout among call center agents, utilizing a pre-trained transformer-based model (BERT) for accurate sentiment analysis and stress detection.

Time Series Forecasting

Hotel Revenue Forecasting

Forecasting global monthly hotel revenue for 2024 using time series and machine learning models.

Recommendation Deep Learning Spotify

Spotify Music Recommender

Machine learning project leveraging the Spotify dataset to build a personalized music recommendation system — exploring content-based filtering, clustering, and deep learning to recommend tracks tailored to user preferences.

Expertise

Skills & Technologies

Tools and frameworks I use to build, deploy, and scale machine learning systems.

Programming

Python C++ C JavaScript

Generative AI / Agents

LLMs Multimodal Models LangChain Prompt Engineering Agent Orchestration Tool Integration

Cloud / Infra

GCP Vertex AI BigQuery Cloud Storage Azure Docker Kubernetes Airflow

Libraries

TensorFlow PyTorch Hugging Face Scikit-learn NumPy Pandas

Tools

Docker Git Spark Flink NGINX Jira

Databases

SQL PostgreSQL MySQL BigQuery

Frameworks

Rasa LangChain Django Flask

Soft Skills

Communication Stakeholder Collaboration Leadership Ownership Agile Development

Education

Academic Background

Strong foundation in computer science and engineering across three degrees.

University of Alberta

Edmonton, AB, Canada

Jan 2021 – Aug 2022

Degree: M.Sc. in Computer Science

Supervisor: Dr. Osmar R. Zaiane

Amirkabir University of Technology

Tehran, Iran

Sep 2017 – Sep 2020

Degree: M.Sc. in Computer Engineering

Kashan University

Kashan, Isfahan, Iran

Sep 2013 – Sep 2017

Degree: B.Sc. in Electrical Engineering

Research

Publications

Peer-reviewed research in NLP, healthcare AI, and distributed systems.

2023 · Conference Paper

Intent and Entity Detection with Data Augmentation for a Mental Health Virtual Assistant Chatbot

A. Zamani, M. Reeson, T. Marshall, M.A. Gharaat, A.L. Foster, J. Noble, O.R. Zaiane

Proceedings of the 23rd ACM International Conference on Intelligent Virtual Agents (IVA)

View paper →

2022 · Journal Article

Developing, Implementing, and Evaluating an Artificial Intelligence–Guided Mental Health Resource Navigation Chatbot for Health Care Workers and Their Families During and Following the COVID-19 Pandemic: Protocol for a Cross-Sectional Study

J.M. Noble, A. Zamani, M.A. Gharaat, D. Merrick, N. Maeda, A.L. Foster, I. Nikolaidis, R. Goud, E. Stroulia, V.I.O. Agyapong, A.J. Greenshaw, S. Lambert, D. Gallson, K. Porter, D. Turner, O. Zaiane

JMIR Research Protocols, 11(7), e33717

View paper →

2022 · MSc Thesis

Developing a Mental Health Virtual Assistance (Chatbot) for Healthcare Workers and their Families

A. Zamani

University of Alberta

View thesis →

2021 · Conference Paper

Developing and Implementing a Mental Health Chatbot to Support Healthcare Workers

A. Zamani, M. Gharayat, J. Nobel, O. Zaiane, E. Stroulia

REMAP-D · Vancouver, British Columbia, Canada

2021 · Journal Article

An Efficient Load Balancing Approach for Service Function Chain Mapping

A. Zamani, B. Bakhshi, S. Sharifian

Computers & Electrical Engineering, 90, 106890

View paper →

2018 · Conference Paper

A Novel Approach for Service Function Chain (SFC) Mapping with Multiple SFC Instances in a Fog-to-Cloud Computing System

A. Zamani, S. Sharifian

4th International Conference on Signal Processing and Intelligent Systems (ICSPIS)

Get in Touch

Let's Connect

Open to collaborations and conversations about agentic AI, RAG, and production ML systems.

Email zamaniali1995@gmail.com LinkedIn linkedin.com/in/zamaniali1995 GitHub github.com/zamaniali1995 Website zamaniali1995.github.io