Skip to content
m_

INDEPENDENT AI CONSULTANCY

AI agents, governance, and applied systems — built to ship.

>_ I design RAG assistantsI design RAG assistants. I build evaluation suites. I write adoption playbooks. I ship full-stack AI products.

I help mission-driven and public-interest teams move from AI hype to working, validated systems — custom agents, evaluation suites, and responsible-adoption frameworks. Strategy and code, from the same person.

geneva / remote / global

Background spanning UN OCHA·Deloitte·P&G·the Munk School of Global Affairs·frontier-AI evaluation programs

01Services

What I do. Four core services + a workshops add-on.

[agents]01

AI Agents & Copilots

From proof-of-concept to validated, production-ready agents. RAG knowledge assistants, advisory/persona agents, and workflow-automation agents that plug into the tools your team already uses (M365, task boards, CRMs, internal data).

  • Use-case discovery & scoping
  • Retrieval-augmented answers grounded in your knowledge base
  • Multi-step / workflow agents
  • A validation suite before anything goes live
  • Handover, documentation & training
[evals]02

AI Evaluation & Quality

If you can't measure it, you can't trust it. I build evaluation datasets, world specifications, and test suites that tell you whether your AI actually works — and keeps working.

  • Evaluation dataset & rubric design
  • World-spec authoring for domain tasks
  • Regression / validation suites
  • Red-teaming & failure analysis
  • Model comparison & selection
[governance]03

Responsible AI Adoption

Practical governance that helps teams adopt agentic AI without the foot-guns. Policies, guardrails, and change management — written for the people who'll actually use the tools.

  • Adoption playbooks (M365 Copilot & enterprise AI)
  • Risk & governance frameworks
  • Data-handling guardrails
  • Staff enablement & exec briefings
[products]04

Full-Stack AI Products

End-to-end web applications powered by LLMs — dashboards, trackers, and internal tools, designed, built, and deployed.

  • Rapid prototyping
  • Next.js / FastAPI + LLM APIs
  • Data pipelines & integrations
  • Deployed on Vercel / Render, then iterated
[workshops]add-on

Workshops & talks

Hands-on sessions on the current AI tool landscape, agent design, and responsible adoption — calibrated to your team's level.

02How I work

A simple path from problem to a working, validated artifact.

01

Discovery (free).

A short call to understand the problem, not just the ask.

02

Scope & proposal.

Fixed deliverables, timeline, and price — no open-ended meters.

03

Build & validate.

A working artifact and the evidence that it works.

04

Handover.

Docs, training, and a clear path to maintain or extend it.

03Selected work

Recent builds — agents, evaluation suites, automations, and a live product.

[live product]live

SitRoom — real-time geopolitical crisis tracker

Context
I wanted a single pane of glass for monitoring global crises — so I built one.
Built
A full-stack app — Next.js/TypeScript front end, FastAPI back end, Claude API for analysis — deployed on Vercel + Render. Built solo, from scratch.
Outcome
A live, public product and my primary portfolio piece.
[agent]

Institutional knowledge assistant

UN humanitarian agency

Context
Staff needed fast, grounded answers from a large body of internal guidance.
Built
A retrieval-augmented assistant grounded on curated institutional source files, deployed inside the org's existing Microsoft 365 / Copilot Studio environment.
Outcome
A working assistant that answers from source material, piloted with the team.

Details on request

[agent]

Scenario-planning agent with a real validation suite

Context
Decision-makers wanted structured scenario analysis they could actually trust.
Built
A scenario-planning agent plus an 18-case validation suite to test its reasoning before use.
Outcome
13/13 pass on the validation cases — shipped with evidence, not vibes.

Details on request

[automation]

Workflow agents that gave a team its time back

Context
Recurring manual reporting and coordination overhead.
Built
A set of workflow agents — a drafting interface over a task board, an automated weekly-update compiler, and a conversational contact-roster tool.
Outcome
Routine reporting and lookups automated; far less manual copy-paste.

Details on request

[evals]

Evaluation dataset for a professional-services AI task

Context
An AI training program needed high-quality, domain-grounded evaluation data.
Built
A world specification and structured evaluation dataset for a consulting-domain task.
Outcome
A reusable spec + dataset feeding model evaluation.

Details on request

[R&D]

Automated trading & market-analysis stack

personal

Built
A Python MCP server for programmatic trading + a market-analysis tool (DuckDB, market-data APIs, Claude).
Outcome
End-to-end capability with live financial data and the modern agent/MCP tooling ecosystem.

Details on request

04About

Strategy and code, from the same person.

My path isn't linear, and that's the point. I started in chemical engineering (University of Waterloo), moved into technology strategy consulting at Deloitte, ran a commercial e-commerce portfolio at P&G, and am now a Master of Global Affairs candidate (Munk School, University of Toronto) with a semester at the Geneva Graduate Institute. Most recently I've been building AI agents and governance tooling for UN OCHA's technology team.

The through-line: I can sit in a room with executives and ship the code that night. I bring the governance and policy depth to do AI responsibly, and the engineering to make it real. I care about AI that serves people — especially in the public-interest and humanitarian space.

Based between Geneva and remote, working with clients globally.

05Engagement

Simple, scoped engagements.

[free]

Discovery call — Free (30 min)

Tell me the problem; I'll tell you whether and how I can help.

Book a discovery call
Start here

Sprint

from $7,500

A scoped 1–3 week build: one agent, one eval suite, or one prototype. Fixed deliverables and price.

 

Project

Custom quote

Larger or multi-part builds, scoped from a discovery call.

 

Advisory retainer

from $2,500/mo

Ongoing guidance and a set block of hours each month.

 

Workshop / talk

from $2,000

Half- or full-day, in person or remote.

Not sure what you need? Start with a free call.

Book a call

06Contact

Let's build something.

Tell me about your project, your team, and what “done” looks like.

LinkedIn
in/mjuzer
GitHub
mojuzer
Location
Geneva / Remote / Global

Prefer email? Write to me directly.