jd /

JOÃO DALLAROSA · FOUNDER

Domain depth.
Applied intelligence.

I turn user needs into AI-powered products, connecting business context, technical decisions and hands-on implementation.

Explore my work Download CV
APPLIED AI / CUSTOMER-FACING IMPLEMENTATION
Financial technology · Conversational AI EU citizen · Open to relocationLuxembourg · Italy · Brazil

THE THROUGH LINE

From the trading desk
to the product decisions.

Eight years across finance and entrepreneurship taught me to connect a technical choice to its operational consequences. Today, I bring that perspective to quantitative systems, conversational AI and the businesses around them.

Selected work

01 — 03 / SYSTEMS & PRODUCTS
01 / FINANCIAL TECHNOLOGYFounder · Jan 2024 — Present

Cop.iaAutomated quantitative investing for retail

Institutional thinking.
Individual access.

I built Cop.ia to bring quantitative and institutional trading approaches to individual investors in Brazil. The product runs a portfolio of 16 automated strategies simultaneously, with trades replicated directly into subscribers’ brokerage accounts through copy trading.

My institutional background became the foundation for a subscription product: strategy research, automated execution and the operations behind a recurring-revenue business.

Quantitative portfoliosCopy tradingPython / MQL5
FROM STRATEGY TO SUBSCRIBER System overview
01020304050607080910111213141516

16 strategies · one quantitative portfolio

AUTOMATED EXECUTIONPortfolio → Copy tradingStrategies execute. Subscriber accounts follow.
XP InvestimentosBTG Pactual

Broker-approved operation in Brazil

16strategies operating simultaneously
Automatedexecution & account replication
>85%annual customer retention
Product, architecture & R&D

END-TO-END RESPONSIBILITY

The system and the business.

I connected strategy development, Python analytics and MetaTrader execution with acquisition, onboarding and subscription operations. Cop.ia operates through XP and BTG — institutions that also shaped my earlier career in finance.

AI-assisted acquisition and onboarding reduced the acquisition cycle by 35%. Annual growth exceeded 20%, with delinquency below 5%.

RESEARCH & DEVELOPMENT

Research that informs the portfolio.

Development includes four-state HMM regime analysis, portfolio risk research and instrument-specific XGBoost experiments. One study covered 39,165 historical trades and 16 features, using independent-data validation to assess practical usefulness.

Python, JSON and MQL5 connect analytical decisions to MetaTrader workflows; APScheduler supports scheduled processing. Experimental scoring remains a research component.

Translate domain expertise into a product people can use.
From strategy logic and execution to subscriber onboarding, retention and unit economics.

02 / AI PRODUCT & MOBILE DEVELOPMENTFounder & Product Creator

LumeConversational AI for emotional support

Built to remember.
Designed for people.

I created Lume from an unmet personal need: a place to talk that could carry a conversation forward. I took the product from idea through app development, user validation and the SEBRAE SC acceleration programme.

The product combines a Flutter mobile app with a Python backend, conversational personas, text and voice, and memory across sessions. My work connected the experience people see to the context, retrieval and cost decisions behind it.

Flutter / DartFastAPIPostgreSQLQdrant
CONTINUITY IN THE CONVERSATION Experience illustration
SSaraCalm & presentlume
A CONVERSATION THAT CONTINUES
FROM LAST TIME

That conversation about work stayed with you.

Context carried between sessions

Do you remember what I told you about Marcos?

I remember. You left that conversation feeling unheard.

Pick up where you left off

Adapted from the supplied Lume website. Illustrative dialogue.

Mobile + AIapp, backend & conversation design
50+users involved in pilot validation
SEBRAE SCbusiness acceleration programme
Explore the product architecture

APPLICATION LAYER

A complete product surface.

The Flutter app uses Riverpod for state management and GoRouter for navigation, with dedicated flows for onboarding, chat, voice and profiles. FastAPI provides the backend, with async SQLAlchemy and PostgreSQL for application data.

Text generation streams through the backend. Voice combines recording, server-side transcription and speech synthesis; usage services track the cost of delivering each interaction.

MEMORY & ORCHESTRATION

Useful context, at the right time.

The backend separates factual memory in PostgreSQL JSONB from retrieval of previous conversations in Qdrant. A search_memory tool lets the agent request relevant history, with user-scoped queries and explicit context budgets.

Custom Python orchestration connects model SDKs, tool calls, session summaries and streaming. This keeps the mobile experience, context lifecycle and consumption tracking within one product architecture.

Flutter + RiverpodFastAPI + model SDKsPostgreSQL + Qdrant

03 / BUILDING WITH AI

AI expands what I can build.
Judgment gives it direction.

I use AI throughout research and development: to explore unfamiliar tools, work through implementation and iterate quickly. I bring the domain context, define the product and stay responsible for the decisions.

THE WORKFLOW
01

Frame the problem.

Translate a user need into requirements, constraints and an architecture worth building.

DOMAIN CONTEXT / PRODUCT SCOPE
02

Build with persistent context.

Give agents defined roles, reusable knowledge and clear handoffs. Use custom tooling and workflow specifications to keep implementation connected to intent.

AGENT WORKFLOWS / KNOWLEDGE SYSTEMS
03

Review. Learn. Iterate.

Evaluate outputs against the original problem, inspect the implementation and feed findings into the next iteration.

HUMAN REVIEW / CONTINUOUS LEARNING
Engineering case studies ↗

Architecture, RAG, containerized services and implementation decisions behind Lume and Cop.ia.

BACKGROUND

João Dallarosa
JOÃO VITOR DALLAROSA LINHARES

Technical curiosity.
Commercial responsibility.

I’m João Vitor Dallarosa Linhares, a founder with an institutional finance background. I work across product architecture, model evaluation, customer discovery and SaaS economics — with responsibility for both the system and the business.

2021 — 2025

Financial Consultant & Educator

Self-employed · Advised and educated 500+ entrepreneurs and investors.

2021 — 2023

Equity Broker / Head of Derivatives Trading Desk

Group XP / BTG Pactual · Institutional derivatives, client relationships and portfolio allocation. Met targets for 17 consecutive months.

2021

SDR · EQI Investimentos

Prospecting, lead qualification and CRM routines. Achieved 110% of the Q1 sales quota.

Education & credentials

Bachelor of Economics · FURB, 2024
ANCORD certification
SEBRAE SC Acceleration Programme · Lume

Technical toolkit

Python · Gaussian HMM · XGBoost
MetaTrader 5 · MQL5 · JSON · APScheduler
Flutter / Dart · FastAPI · PostgreSQL
Qdrant · RAG · Docker · Agent workflows

Languages & location

Portuguese · Native
English · Fluent
Santa Catarina, Brazil · Open to EU relocation

OPEN TO AI & TECH OPPORTUNITIES IN THE EU

Let’s build something
that holds up in practice.

Applied AI · Technical product · Financial technology

Download CV LinkedIn

Citizenships: Luxembourg, Italy & Brazil · EU citizen