PRODUCT · AI BUILDER · CUSTOMER CENTRIC
Product & AI Builder — I ship things that solve real problems and measure whether they worked.
I'm targeting product management, strategy, and AI-forward roles where I can work on hard problems, ship fast, and keep learning.
A small prototype I built based on a Shopify shipping setup assignment prompt. The problem: many merchants abandon setup before finishing when they hit unfamiliar shipping terms. My solution: a guided assistant that asks two plain-English questions and configures the setup flow automatically. Built and deployed as an interactive prototype on Netlify.
Led a multidisciplinary team in designing an AI-powered assistive system supporting visually impaired individuals and dementia patients with spatial awareness and memory reinforcement.
Building an end-to-end LLM + RAG system that dynamically retrieves role-specific knowledge and generates structured interview simulations with feedback.
Engineered and benchmarked PaDiM, SPADE, and CLIP-based models using PyTorch and OpenCV for video anomaly detection.
I identify problems, build solutions, and measure whether they worked. My background spans AI engineering, product delivery, and data analysis — I use all three to ship things that actually move the needle. My work spans prompt engineering, retrieval pipelines, backend APIs, and enterprise-grade deployment.
I enjoy converting ambiguous product and policy constraints into deployable systems that teams can trust.
Enabled reliable, structured AI-assisted case review decisions in the SAMS enterprise case management system by validating model responses for accuracy — engineered using Jinja-based prompt templates for production LLM workflows.
Reduced manual case-processing overhead in a regulated government environment by approximately 79% by building secure RESTful APIs and backend orchestration layers for agentic, AI-assisted decision workflows.
Improved reliability and stakeholder alignment of LLM integrations by defining measurable success metrics and iterating on prompt/retrieval pipeline design — deployed and debugged via Azure DevOps CI/CD.
Improved targeted marketing effectiveness by 15% by translating ambiguous business questions into structured, metric-driven insights — analyzed customer behavior and market segmentation data using Python and Tableau.
Enabled faster, clearer decision-making for cross-functional stakeholders by distilling complex data into actionable recommendations — presented through dynamic dashboards.
Improved data collection efficiency and downstream analysis quality by 20% by designing and deploying customer surveys with optimized question logic.
Streamlined cross-department operations by integrating backend services and automating workflows — implemented and configured using ERPNext modules.
Improved reporting accuracy and operational efficiency by reducing manual processing by approximately 30% — achieved by building automation pipelines.
Featured reads from my Substack on practical AI concepts and real-world thinking.
You Need Canadian Experience to Get Canadian Experience. Here's What I am Doing
Read on Substack
In 2020, Google fired Timnit Gebru for telling the truth about AI. In 2026, the women raising the alarm inside the biggest AI companies are still getting shown the door.
Read on Substack
The AI tools you open every day are sharing more than you think — and the one law designed to protect you just got killed before it started.
Read on Substack
(and the Tools Worth Using in 2026)
Read on Substack
How to Get a Job With a Computer Science Degree in 2026
Read on Substack
Your weekly curated reading list — dropping every Friday.
Read on SubstackPick role + difficulty, beat the timer, stack a streak, and train like an interview game.
A churn dashboard spikes for one cohort. What is your first 30-minute investigation plan?