Ranbeer Makin

AI Educator, 19+years of experience in software engineering field

Jev 101 - System One Model

Understanding of System One Model

LLM vs Jev

Jev examples with code and nocode walkthrough

Jev use cases

AI first PM: Tech 101, Claude mastery, AI agents, Vibecoding and Evals.

Lead PM @ Booking.com, PhD CS

Vladimir Kalmykov 

PhD Computer Science, first Tech PM in Booking.com, author @ ProductDo.io 

Sr. PM ML @ Booking.com, MBA

Andrew Mende

Sr ML Product Manager Booking.com, MBA

Cohort 10

Cohort Nov'26

✅ Why invest in learning with us?<ul><li>We’re Lead PMs at Booking.com, each with 10+ years of product experience</li><li>We have a rule: we teach only what we actually use in our day-to-day product work, with a strong focus on practice over theory</li><li>More than 14,000 students have already learned through our hands-on, “learn by doing” approach in ProductDo.io Simulators</li></ul>✅ Modern AI-first PM:<ul><li>Use AI agents (e.g. Claude Code) to accelerate the routine (sprints, analytics, PRDs, etc)</li><li>Can easily vibecode a UI feature, and then connect it to real APIs</li><li>Can set robust AI Evals to judge the quality of LLM features in their product</li><li>Knows the basics of Tech (APIs, services, tools, MCPs) to confidently lead their tech team</li></ul>✅ What makes this course different?The course is 80% hands-on because that’s how real skills stick. Don't be scared - you are a hand-held all the way. We simulated a startup for you with:<ul><li>real Jira backlog that you analyze with Claude</li><li>real codebase on top of which you vibecode new features</li><li>real customer support chats that you need to build evals for</li><li>and more</li></ul>You're the main character - a PM who solves product tasks. Check out the Free Resource on Evals below to get the feel of the flow.

Learn by doing in a simulated start-up experience. From acting Lead PMs.

Gain tech fluency to speak the same language with your dev / AI team 

Learn how to use Claude to delegate routine tasks and build custom agents

Add LLM features, measured by evals and LLM-judges

Start vibe coding confidently to turn ideas into working software

The course is designed to explain all the topics from scratch. 

No prior tech or AI experience needed

You need Claude Pro (20$) to do exercises from the course. 

Claude Pro for 1 month

Senior Product Managers who want to level up their skills with new tech and AI skills

Jr/Middle Product Managers who want to lead with confidence and accelerate their growth

Program/Project Managers, Producers, Designers, Data Analysts, because tech and AI-topics are the same for everyone (see reviews)

Make more cohesive product and design decisions at AI speed with clearer logic, stronger value cases, and better judgment under pressure.

Product & Design Leader at Owtcome

Owtcome

Krasi Bozhinkova

Cohort - May 2026

Cohort - Nov 2026

AI has made building faster, but product success hasn't gotten easier.Teams are now making more product-shaping calls with less time, more pressure, and incomplete evidence. The challenge is no longer just whether a single decision is right, but whether many fast decisions still add up to a coherent product direction.This course teaches a repeatable Product Judgment System for that challenge: how to clarify the critical user need that should guide decisions, build the value case, define the trust conditions, find where growth breaks, and choose the right product move when conditions change.It is built for PMs, designers, and product leaders whose product thinking needs to hold up when AI creates more ideas, more prototypes, more pressure, and less time to wait for certainty.You’ll leave with practical decision tools to apply judgment where it matters most: choose which user need deserves focus, pressure-test weak value cases, decide what deserves momentum, protect trust, and know what must change when the market shifts.The goal is to help you keep product decisions coherent under pressure and prioritize the work that will drive adoption, retention, trust, and business outcomes.

Apply a Product Judgment System for better decisions at AI speed

Create clarity before weak logic becomes a shipped product

Build the value case when the evidence is still incomplete

Define the trust conditions that keep product decisions coherent

Fix what blocks growth and choose the right response

PMs, designers, and cross-functional leaders with 3-5 years of experience, working through ambiguity, trade-offs, and business outcomes.

Level of experience

Bring one live product, feature, or strategic decision to apply the course directly to your own context.


What to prepare

This course is designed for practitioners who want to make better product decisions in AI-shaped markets, not to teach a specific AI tool.

No technical AI background required

Mid-to-senior PMs who need to make harder product decisions fast and connect the value case to adoption, trust, and growth.

Designers expanding their influence who need sharper language to connect design decisions to product and business outcomes.

Product-minded engineers and cross-functional leads stepping into ownership who need to make hard product calls with incomplete evidence.

By the end you'll have a benchmark for your job, a small model measured against a big one, and a cost sheet that says what it's worth.

Developer Relations, PrismML

Jai Bhagat

Owns a GB10, has spent $4,000 renting models, and prices AI work live.

Expertise into products people pay for

Nicole Mercede

Went from billing hours to owning what she produces; runs the pitch week.

Cohort 1

If you do knowledge work for a living, a large share of your week is one judgment call made over and over with slightly different inputs: triage the inbox, categorize the request, decide how the company should answer, write it in the house voice, log it, report on it. Nicole used to bill that kind of work at $85 an hour for about twenty-five hours a week, roughly eight thousand dollars a month of labor that a general-purpose model can already do most of, and that a small specialized model might do better and cheaper. This cohort exists to find out whether that's true for your job, on hardware you control, and what it would be worth if it is.I'll be straight about what this is: an experiment with a method, not a guaranteed outcome. I own an NVIDIA GB10 and I've spent about $4,000 of my own money renting frontier models, so I know what each rung costs, but I don't yet know how far a small model trained on your domain gets on Apple Silicon or the GB10. We find out together over three Fridays, with a benchmark that's allowed to tell us no. If the answer is no, you leave knowing it with the numbers attached, which is a result I'd pay for.

Your expertise is worth more as a model than as hours

Build a benchmark for your own job

Train a small model and read its result honestly

Pick your bet and price it

Leave with a pitch or a decision

Bring one job you know well and a laptop; most recent ones can run a small model, and I run the heavier jobs on my GB10.

No GB10, and no engineering degree, required

Knowledge workers whose week is mostly one repeatable judgment, and who would rather sell the model than keep selling the hours.

Founders and domain experts who know a corner of healthcare, education, or public infrastructure better than any model does.

Builders who can already run and fine-tune a model but have never had to defend its unit economics or name who would pay for it.

By the end, you can apply a robust and replicable evaluation method to any AI feature, as the PM who manages it or the Builder who ships it

Ex-PM turned AI-Native Product Builder

Instacart

Amazon

PayPal

Optimizely

GoDaddy

Jules Boiteux

Former Senior Product Manager, now founder of Vibe Coding Academy & AI educator

Q4 2026

🚨 Save 20% <a target="_blank" rel="noopener noreferrer nofollow" class="PlaygroundEditorTheme__link" href="https://maven.com/jules-boiteux/building-your-first-ai-eval?promoCode=FIRST-AI-EVAL">with this link</a>----💰 The most hands-on AI evals course you'll find under $500----AI evals are the hottest skill an AI-native PM can have: more and more features run on an LLM, and their accuracy keeps competitors at bay.The problem is where to start. Evals feel like an engineer's job, the information is scattered, the courses long and expensive.Concrete practice, not slides. You run every eval on a live example: an agent sorting the real feature requests of <a target="_blank" rel="noopener noreferrer nofollow" href="https://excalidraw.com/">Excalidraw</a>, an open-source whiteboard app.In 3 sessions of 1.5 hours, you go from zero to your own eval tool: a golden set, a scored system prompt, an LLM as a judge, then a tool you vibe code to run your AI evals.No coding, no Claude Code knowledge required: the method works with any tool (Codex, Cursor...). I teach with Claude Code, the easiest way to apply them. We set up together in session 1.----🛠️ 60% hands-on practice, 40% theory📚 Lifetime access to recordings and future cohorts✍️ Written review of your golden set and judge🧪 A concrete example to replicate for any AI eval you run🗂️ A GitHub boilerplate for your own evals👥 A private Slack channel, even after the course

The course for AI-native PMs who want to build their first AI eval from scratch

How to build a golden set of data you can trust

How to make your setup ready for AI evals

Learning the basics around AI evals

How to calibrate an LLM judge

How to vibe code your own eval tool with Claude Code

How to apply the method to any AI feature you work on

Taught with Claude Code, so Claude Pro is smoothest; Codex or Cursor Pro work too. Plus a laptop you can install on.

A subscription to Claude, OpenAI or Cursor

You have used Claude or ChatGPT and know what a prompt and an API are. No coding: session 3 is vibe coding, you direct.

Basic AI literacy (no coding required)

We run every eval on Excalidraw's public issues, taught as a method: each step maps to your own AI feature, when you are ready. 

No AI feature of your own needed

Product Managers, Product Owners and AI-native Product Builders who ship AI features and want proof that a change helped

Software engineers building LLM features who want a score next to every prompt version instead of a gut feeling

Leaders who steer product and tech teams on AI features, who want to talk evals with them, challenge on evidence and set priorities.

Agentic AI

Coding with AI

AI Workflows

Claude Code

OpenClaw

Vibe Coding

AI Evals

AI Transformation

RAG & Search

AI for PMs

AI for Engineers

AI for Designers

AI for Marketers

AI for Founders

Product Sense

Product Discovery

User Research

Prototyping

Growth

Analytics

Tech Foundations

Strategy

Influence

Leadership

Career Growth

Machine Learning

LLM Ops

Context Eng

Security

System Design

Figma

Design Systems

Visual Design

Design Strategy

Automation

Content Marketing

Demand Gen

Product Marketing

Positioning

Social Media

Brand

B2B Marketing

SEO & AEO

AI for Leaders

AI Governance

Communication

Management

People Operations

Exec Presence

Storytelling

Goal-setting

Personal Brand

AI for GTM

Sales

Customer Success

GTM Engineering

Everyone

Founders

Operators

Data Scientists

Business Analysts

User Researchers

Project Managers

HR Professionals

Sales People

Lawyers

Finance

Investors

Real Estate

Educators

Creators

Product managers who drive strategy, development, and launch of products.

Engineers and engineering leaders who build, deploy, and scale software systems.

Applying AI tools and techniques to professional workflows — assign alongside more specific AI topics.

Automating professional workflows with AI tools — productivity, operations, and task automation.

Jev 101 - System One Model

What you'll learn

Why this topic matters

You'll learn from

Ranbeer Makin