Mission log / selected work

Products are systems.
So is the work.

Four examples of product leadership under real constraints: aligning people, designing the decision loop, and creating conditions for trustworthy adoption.

Enterprise AI Personalization Platform

01 / Platform · GenAI
Operating context

Enterprise SaaS teams needed real-time customer insight and personalized content at scale, while legacy workflows could not introduce GenAI without compromising compliance and reliability.

Leadership brief

Architected the platform with GenAI and vector databases; owned the end-to-end model lifecycle and MLOps on AWS/Azure; led 5–8 PMs across a multi-workstream roadmap.

System outcome

Deployed production GenAI for real-time insights, recommendations, and sales automation, with an observability stack that improved reliability and explainability at scale.

Decision log: Sequenced by customer impact and feasibility—not model novelty. The highest-trust use cases came first, turning compliance and business stakeholders into advocates rather than blockers.

AI-Powered Document Generation Tool

02 / Workflow · GenAI
Operating context

Investment directors spent 3–4 hours each week drafting portfolio commentaries. The repetitive work constrained reporting cycles, while accuracy and compliance could not be optional.

Leadership brief

Led LLM evaluation and selection, defined the prompt architecture for personalized commentary, and aligned compliance and front-office users around the operating model.

System outcome

Drafting time fell 80%. More than 40 investment directors adopted the tool, and compliance approved the first pass because guardrails were part of the product spec from day one.

Decision log: Selected a smaller, fine-tuned model over GPT-4 to stay within compliance data boundaries—the right capability-to-trust tradeoff for this workflow.

Real-Time Detection & Prevention System

03 / Real-time · ML
Operating context

A financial-services platform faced rising fraud loss while rule-based detection missed novel attack patterns. High false positives were also putting legitimate customers through needless friction.

Leadership brief

Owned product strategy from model selection and feature-engineering priorities through the risk thresholds that balanced fraud prevention against customer experience.

System outcome

False positives fell 35%, improving customer satisfaction and reducing the manual-review backlog while the system addressed novel fraud patterns.

Decision log: Added explainability because fraud-operations teams needed to understand the signal before they could operationalize it. Adoption—not only regulatory compliance—depended on it.

Intelligence & Alert Prioritization Platform

04 / Prioritization · AI
Operating context

Security teams were drowning in alerts and spending more time triaging noise than responding to material threats. A reactive posture was leaving organizations exposed.

Leadership brief

Defined the platform vision and roadmap, led AI-driven threat-scoring requirements, and worked directly with analysts to design the investigation workflow.

System outcome

Mean time to detect fell 60%. Analysts faced less alert fatigue and could move from reactive triage toward proactive threat hunting.

Decision log: Excluded automated response from v1. Getting threat scoring right was the harder and more valuable problem; automation built on poor signal only makes the wrong move faster.

Product lab / side project

Young Tycoon — Sweet Street

A small experiment built for curiosity, play, and the fun of making an idea real.

Game experiment · Ages 8–10
The spark

Could a simple game give kids a low-stakes place to practice pricing, saving, reinvestment, leadership, and trade-offs?

What I explored

A kid-first game loop, 14-day progression, rewards, parent-facing context, and feedback moments—turned into a playable prototype with AI-assisted building tools.

Where it is now

The prototype is live for families to try. It is a side project and learning experiment, not part of my professional work or a claim of validated educational impact.

Design idea: Put the lesson inside the choice. Players buy inventory, set a price, serve customers, and decide whether to save or reinvest, then see the consequences through the shop economy.

Play Young Tycoon ↗ Public prototype · No Lovable account required