Hey thereHolaCiao,

I’m Paolo

MBA at MIT Sloan, ex-BCG.
I scope products, then build them myself.

See what I’ve built
Portrait of Paolo Fabia smiling, in a navy quarter-zip
Open to PM internships · Summer 2027
$8M+ revenue uplift from one product launch
40+ GenAI use cases, design to production
5 years of work experience across data analysis, product and strategy

How I got here

2016 to Now
  1. 2016

    BSc Industrial Engineering
    MSc Computer Science

    • Graduated in the top 4% of the class; sole recipient of the Graduation Award for leadership and academic record in a class of 800.
    • Led a 5-student team as product manager for the capstone course: the insurtech policy aggregator in Things I’ve built, designed in Figma and built full-stack with React and FastAPI.
    • President of the Engineering Students Council, representing 5,900+ students, driving communication with faculty leadership and overseeing project execution across student-led initiatives.
  2. 2021

    Data Analyst

    • Developed AI/ML models assessing dropout risk for 1.2M students in Chilean public education, with the documentation to put them in stakeholders’ hands.
    • Prototyped automated monitoring of Regional Education Plans covering 5,000 establishments.
  3. 2022

    Strategy & Dev. Analyst

    • Ran the end-to-end product lifecycle for an internal market-mapping tool, writing requirements for 2 engineers to profile 1,500+ U.S. insurers by location, size and portfolio. The go-to-market team adopted the output as its primary input for market-entry strategy.
    • Built the financial models behind the U.S. expansion business plan, prioritising 10 product initiatives by investment need.
  4. 2023

    Consultant

    • Designed 40+ GenAI use cases for a Latin American retail chain — CrewAI agentic workflows, GPT-4o content generation, Rasa chatbots — and wrote the innovation roadmap that took them into production with operations and CX.
    • Defined product strategy for a consumer-goods group (app features, data architecture, real-time reporting); the launch drove $8M+ in revenue uplift over three years.
    • Built an 18-month IT roadmap for a two-bank merger, integrating 250+ systems against current-state processes, future priorities and emerging AI capability.
    • Redesigned an IT operating model and cut procurement spend by $30M for a large Mexican retailer; a parallel redesign with 10+ managers delivered $4M for a real-estate firm.
    • Researched digital-banking adoption across 5 countries and 100+ customers to set requirements for digital workflows.
  5. 2026 · Now

    MBA Candidate

    • Pursuing the Product Management Certificate, with coursework in Product Strategy, Product Marketing, Leading Agentic AI and Product Management Lab.
    • VP of Content for MBA Chile — organising the annual conference for 300+ Chilean MBA students in the US, April 2–3, 2027.
    • Member of the Product Management Club, Technology Club, Ski Club and Sloan Pride Club.

Things I’ve built

Other relevant projects at work & school

ERP analytics dashboard

BCG · 2024

Power BI reporting wired into SAP for an energy company.

90+users adopted it

Insurtech policy aggregator

UC Chile · 2021

Every insurance policy in one dashboard, duplicates flagged.

PMof a 5-person team

Telegram expense bot

Personal

Snap a receipt; the bot drafts the expense.

Dailyin my own use

Built with

  • React
  • Next.js
  • Supabase
  • Python
  • TypeScript
  • Figma
  • FastAPI
  • Claude
  • Lovable
  • SAP
  • Power BI
  • Telegram
  • PostgreSQL
  • Vercel
Paolo on a dock next to an MIT dinghy with a red sail, Boston skyline behind
On the Charles · MIT Sailing

Beyond work

  • Competitive swimmer at national and South American championships. Fifteen years of 5 a.m. training is where I learned that consistency beats intensity — the habit I’ve leaned on hardest everywhere since.

    2008 — 2023

    Paolo with his swim team on the pool deck
    Paolo swimming backstroke in a race
  • Fundación Botín fellowship

    Public leadership · Spain

    One of 20 selected from 6,000 applicants across Latin America.

    Fundación Botín fellows holding Latin American flags in front of the Santiago de Compostela cathedral

    Trabajos de Invierno

    Volunteer · UC Engineering

    Winter volunteer trips, helping low-income families in southern Chile with construction.

    Paolo and a fellow volunteer, muddy after a day of building
    The volunteer team in front of the wooden frame of a house

    El Barrio

    Co-founder · Santiago · 2020

    A handful of us organised emergency aid for low-income families when everything shut down. Nothing teaches logistics like doing it with no budget and no time.

    El Barrio team card: Paolo Fabia, in charge of finance
    Paolo, wearing a mask, with a woman at the door of her home
  • Andes first, New England now. I joined the Sloan Ski Club before I’d unpacked — and yes, I once built a trek-picker app partly because I wanted to know which trips I could actually afford.

    Sloan Ski Club

    Paolo in ski gear, sitting on a bench in the snow
    Paolo and friends resting with their skis on a snowy slope
  • President, Engineering Students Council

    UC Chile

    Represented nearly 6,000 engineering students, worked directly with faculty leadership and oversaw a wide range of student-led initiatives.

    Paolo speaking at a podium at a 2019 university event
    Paolo speaking at a university ceremony, shown on the big screen

    VP of Content, MBA Chile

    Conference · April 2–3, 2027

    Organising the annual conference for 300+ Chilean MBA students in the US.

Let’s talk.

Hiring a product intern for summer 2027? Write to me in English, español or italiano.

Live · 2026 · Solo build

Collide

Matchmaking for the room you’re already in.

Guests at a conference or mixer scan a QR code, answer five short questions and get a few introductions, each with a reason to talk, an opening line, a time and a place. No app to install, no account to create.

Used at events of 300+ attendees, with 400 matches suggested.

Visit lets-collide.com (opens in a new tab)

The call I made
A good match is one shared thread plus one useful difference. The AI scores every pair on exactly that — something in common, and something one can give that the other needs — and marks down near-twins as hard as strangers. It reads answers only, never names or contact details, and when it fails nothing is made up: the organiser gets a retry.
What I left out
Accounts. Guests join with a first name and organisers with a passcode. Email and LinkedIn are shared only after both people tap “We met”, so joining costs ninety seconds and nothing changes hands before a real conversation has happened.
The risk I’m carrying
Load. Pages check for updates every four seconds instead of holding a live connection. That kept the build simple, and it means a 100-person event makes about 90,000 requests an hour — the first thing to replace as the rooms get bigger.

Live · With a teammate

Sloanies4Sloanies

Honest reviews of every MIT Sloan class.

A student-run web app that gives every MIT Sloan course one page: the official catalog data next to what the students who took it report — workload, project type, travel and what it took to bid in. The evaluations give you the score; this tells you what the score means.

Live with 250+ users and a catalog of 400 Sloan courses.

Visit rate-my-sloan.vercel.app (opens in a new tab)

The call we made
Nothing is written for a student. The first design had AI turn free text into fields, and we cut it. A review goes up exactly as its author wrote it, after a tone check that blocks insults — no summaries, no filled-in fields, and no seeded reviews to make an empty page look busy.
The trade-off
Give one to see all. Past the six classes on the home page, you need MIT sign-in and one review of your own. It turns casual browsers away on purpose: a course page is only worth reading once enough people have written on it, so reading is paid for in reviews.
The risk we’re carrying
Trust. Ratings and bids are self-reported and unverified, and the site says so above every list of them. A single report can look like a fact, so every number shows how many reports stand behind it.

Live demo · Hackathon build

The Missing Aisle

The snacks from home U.S. stores don’t carry.

A marketplace inside the U.S. for the snacks, spices and gifts from home that U.S. stores don’t carry. Small importers list in their own language and reach everyone looking; buyers search by name, by a photo of the package, or by describing it when they don’t know what it’s called — and request what nobody sells yet.

Built at a Lovable hackathon: a demo catalog of 120 products from 30+ countries.

Visit missing-aisle.lovable.app (opens in a new tab)

The call I made
A failed search is demand, not a dead end. When a description matches nothing exactly, the AI shows the closest products and drafts a request in the buyer’s own words, ready to edit and post. Every miss becomes something a seller can see and answer.
Labels that don’t overclaim
Origin is marked “seller-declared”. A Firecrawl check of Walmart, Target, Kroger, Costco and Whole Foods says whether a product already has U.S. listings, and it is labelled as availability, not authenticity, because a search result proves neither local stock nor where something came from.
The risk I’m carrying
The cold start. A marketplace needs sellers to draw buyers and buyers to draw sellers, and today’s catalog is a demo. The request board is the bet: a +1 count shows an importer what to bring before they list anything.

BCG · 2024 · Owner

ERP analytics dashboard

Power BI reporting wired into SAP for an energy company.

A Power BI reporting layer wired straight into SAP for an energy company, replacing manual reporting across finance and operations. I owned the build: prioritised functional and technical requirements with both teams and directed delivery.

Adopted by 90+ users and 6 managers; directed a 3-person team.

The call I made
Operations went first. They carried the daily volume and lost the most hours to manual reporting, so their views shipped ahead of finance’s — and finance stayed on spreadsheets until the second round.
The pushback I absorbed
The numbers were the argument, not the tool. The new layer disagreed with the spreadsheets people had been reconciling by hand, and until every difference was explained it was a second opinion rather than the source of truth.
What I’d do differently
I treated the build as the hard part. It wasn’t — getting 90+ people to change how they work was, and I gave that almost no room in the plan.

UC Chile · 2021 · Product manager + build

Insurtech policy aggregator

Every insurance policy in one dashboard, duplicates flagged.

A consumer platform that pulls a person’s policies from every provider into one dashboard, surfaces duplicate coverage and savings opportunities, and lets them contact insurers directly. Capstone project at PUC Chile.

Led a 5-person team as product manager — designed it in Figma, built it full-stack.

The call I made
A couple of providers were wired for real and the rest were mocked. No insurer was going to open its data to a student project inside one semester, so we proved the whole flow end to end where we had live integrations and simulated the rest.
The pushback I absorbed
It came from inside the team. People wanted to add features, and with a fixed deadline my job was to keep saying no to good ideas — which is harder to do to five classmates than to a client.
What I’d do differently
We barely used the expertise around us — teaching assistants, professors, alumni who had already solved these problems. Asking earlier would have unblocked work that instead slipped into the weeks when everyone’s coursework peaked, and the team carried that as pressure.

Personal tool · In daily use

Telegram expense bot

Snap a receipt; the bot drafts the expense.

I photograph a receipt, send it to a Telegram bot, and it reads the receipt and creates the draft in my company’s reimbursement system. All that’s left for me is hitting submit.

Turned a recurring admin chore into one tap — built for myself, in use ever since.

The call I made
It stops at the draft. The bot fills everything in and then waits, because a miscategorised expense filed automatically becomes a conversation with finance, and two seconds of review costs less than that. I automated the typing, not the judgement.
What I’d do differently
It fails silently. No tests and no error handling, so when a receipt confuses it or the API is down I find out later, going to look for a draft that was never created. One “I couldn’t read this” message back into the chat was almost free, and I skipped it.