Samiul Islam · Senior AI Product Manager · Toronto

I design and build agentic AI systems. 

13 years in controls engineering and IoT before product. Now I ship multi-agent orchestration, MCP tooling layers, and voice AI.

See the workGitHub
years controls & IoT
13
years controls & IoT
enterprise product teams
8
enterprise product teams
years shipping
15+
years shipping

Featured work

Prototypes as proof.

Public builds, end to end - product thesis, architecture, evals, and the tradeoffs written down.

TypeScript / Next.js / Claude Agent SDK

Meridian

A PM agent that turns raw inputs - tickets, transcripts, docs - into decision-ready artifacts: PRDs, user stories, release notes. Approval-gated external writes, streaming UI, and an evaluation harness (15 cases, 57 structural checks) so no skill ships unverified.

Python / LangGraph / Next.js

HealthSignal

A personal health engine that turns Garmin wearable data into a daily Wellness Score. A parallel council of experts - sleep, performance, stress agents - synthesized into one narrative.

Design tokens / Tailwind preset

japandi-starter

A portable Japandi design system: tokens, a Tailwind preset, and a plain-words style guide that AI coding tools can follow on any stack.

Live demo

Model Context Protocol

MCP tooling

Building with and on MCP: a production MCP + multi-agent system at work, and open-source servers like Freshdesk-MCP for AI-driven support operations.

Swift / macOS

AppSeed

Wraps any web app as a native macOS .app.

The through-line

From factory floors to agentic AI.

13 years

Controls engineering

PLCs and industrial IoT across automotive, mining, and oil sands. The engineering depth is the point: I validate ideas hands-on before they reach production engineering.

Formula.Monks

Enterprise AI product strategy

AI product strategy across 8 product teams, shipping enterprise AI products for Fortune 500 clients on an agentic AI framework over a semantic tool layer.

Thinkrr

Senior AI Product Manager

Shipping a voice AI platform and a unified MCP + multi-agent system.

Speaking

On stage around Toronto.

Sauga Tech · June 2026

Defending the Prompt

An AI security talk on prompt injection and what it takes to defend agentic systems in production.

Talk resources

Regular speaker on agentic AI, evaluation harnesses, and AI security at Toronto-area meetups - Sauga Tech, Mississauga .NET, and the local AI scene.

Elsewhere

Find me where the work is public.

LinkedInGitHub