Edson Cilos Vargas Júnior
Verified Expert in Engineering
Python Developer
Florianópolis - State of Santa Catarina, Brazil
Toptal member since July 27, 2022
Edson has a strong background in building and validating machine learning models and AI systems. Curious, motivated, and diligent, he is experienced in finance and familiar with quantitative reporting. He has experience in applied AI projects, including large language model (LLM) agent infrastructure and knowledge platforms. Edson's passion is learning and sharing knowledge, bringing value to companies. He is enthusiastic about reaching new heights in intelligent systems.
Portfolio
Experience
- Python - 7 years
- Backtesting Trading Strategies - 7 years
- Machine Learning - 7 years
- Quantitative Modeling - 6 years
- Trading Systems - 4 years
- Algorithmic Trading - 3 years
- Risk Models - 2 years
- Julia - 2 years
Preferred Environment
Agent-based Modeling, Algorithmic Trading, Risk Models
The most amazing...
...thing I built was a high-performance agent-based simulator that models thousands of traders trading concurrently across multiple markets.
Work Experience
Quant Researcher & Developer
Scilabs Solucoes Tecnologicas em Financas
- Architected an MCP-native knowledge platform on PostgreSQL that exposes a semantic store to large language model (LLM) agents, with hybrid retrieval, provenance traversal, and a signal analyzer for detecting gaps in projects and other AI systems.
- Designed a hybrid retrieval layer for the platform combining pgvector dense search and PostgreSQL lexical search, fused via Reciprocal Rank Fusion over model-versioned embeddings.
- Built a resilient, multi-chain RPC pipeline that consolidates swaps, liquidity, TVL, and other on-chain metrics from 100+ DEXs, powering dashboards and reports for trading and risk mitigation.
- Engineered a high-performance agent-based simulator that models thousands of traders trading concurrently across multiple perpetual-DEX markets.
- Delivered end-to-end consulting on crypto trading-bot development and deployment, plus risk dashboards covering PnL vs protocol, open interest, market skew, among others.
Professor
The Federal University of Santa Catarina
- Designed and delivered AI lectures for mathematicians and scientists at UFSC, connecting mathematical foundations to modern AI practice.
- Built a machine learning (ML) pipeline for ATR-FTIR microplastic classification: 91% test accuracy, 94% on an independent dataset with no re-tuning. Corrected snooping bias in prior work.
- Taught and mentored 200+ students in state-of-the-art machine learning, guiding 30+ hands-on projects spanning spam filtering, digit recognition, sentiment analysis, XGBoost, Gaussian mixture models, and wind-turbine failure analysis.
Head of Risk
Zaros Inc
- Modeled an artificial financial market using agent-based simulation, capable of simulating leveraged operations and interactions among various agents in the financial market.
- Acted as a co-author of an article that described the functioning of a new decentralized protocol for perpetual futures. In partnership with the company's head of blockchain, I proposed several improvements to the project.
- Became one of the principal authors of the smart funding algorithm proposal, a dynamic funding rate updated according to market volatility, asset risk, and open interest imbalance.
- Acted as the principal author of the eClusters proposal, which employs hierarchical risk parity (HPR) to group asset liquidity into clusters, solving the liquidity fragmentation in decentralized perpetual futures protocols.
Senior Treasury Analyst
Foxbit
- Optimized and enhanced strategies in HummingBot, including improvements in cross-exchange market making, resulting in reduced exposure and significantly increased results.
- Deployed, maintained, and monitored bots for market making, providing liquidity to order books of various cryptocurrencies at a meager cost.
- Developed an integrated financial reporting system aggregating historical data from multiple cryptocurrency exchanges (Binance, Foxbit, OKX, Bitstamp, Bitso, etc.), banks, and other business partners.
- Collected data via API or web scraping, transformed, and stored in a database. Through a telegram request, a consolidated view of the results from different bots and company stained.
- Developed a project using Agile methodology with boards in Jira. Managed the treasury team in charge of the financial reports for the operations of the bots and OTC.
Quant Developer
Truss Investments
- Developed a backtesting system to evaluate stock market strategies and produce automated reports.
- Implemented a backtesting system using Julia, up to ten times faster than the open source Python libraries.
- Produced reports and analyzed stock market data, including evaluating strategies to operate on day trading.
Math and Data Science Expert
Leve
- Discussed theoretical and practical issues of the mathematical models used in the company's grant credit system. The designed model will be used for several years and positively influence the risk vs. return.
- Analyzed dismissal data from CAGED and RAIS databases, while the exploratory data analysis generated important insights to help Copula's model used in risk modeling.
- Guided the newly hired mathematician to ensure that the company absorbed the shared knowledge.
- Contributed to the company's projects by suggesting the best tools and frameworks.
Quant Developer
PutCallBot
- Developed new stock market strategies using machine learning models, including neural networks and principal components analysis (PCA).
- Automated generation and processing of broker notes. The system could dynamically read a PDF and store the main contents in an SQL database, with more than 250 pages of broker notes in less than 15 minutes.
- Developed a simple admin webpage to upload PDF broker notes or request a specific user's PDF broker.
Experience
Machine Learning Course Development
OVERVIEW
• Machine learning fundamentals
• End-to-end machine learning project
• Machine learning projects development using Python
• Best practices in machine learning projects
• Machine learning mathematics made easy
• Extra in-depth materials
Algorithm Consulting for Facial Recognition
• Principal component analysis
• Linear Discriminant Analysis
• Laplacian Discriminant Analysis and variations of such algorithms
The goal was to develop in-house knowledge about dimensionality reduction and apply such algorithms to computer vision, specifically facial recognition. It is important to note that even the lowest level algorithms, such as methods for computing eigenvalues, have been implemented from scratch. Also, spent a reasonable time reading scientific articles to support the project's main findings.
LMS Platform
Education
PhD in Mathematics
The Federal University of Santa Catarina - Florianópolis, Brazil
Visiting Scholar in Mathematics
University of Konstanz - Konstanz, Germany
Bachelor's Degree in Mathematics and Scientific Computing
The Federal University of Santa Catarina - Florianópolis, Brazil
Skills
Libraries/APIs
Scikit-learn, Binance API, Pandas, NumPy, XGBoost, REST APIs, API Development, OKX API, TensorFlow, Keras, Telegram Bot API
Tools
Claude, Claude Code, LaTeX, MATLAB, Jupyter, Microsoft Excel, BigQuery, GitHub, Google Compute Engine (GCE), Amazon SageMaker, LearnDash, SendinBlue, Docker Compose, Jira, Plotly, Looker
Languages
Python, Julia, SQL, Python 3, GraphQL, C#, JavaScript, HTML, CSS, Solidity, C++
Frameworks
Agentic Frameworks
Paradigms
ETL, Quantitative Research, Model Context Protocol (MCP), Agent-based Modeling, Database Design, Web UI Design, Automation, Agile
Platforms
Linux, Jupyter Notebook, Blockchain, Ethereum, Decentralized Exchanges (DEXs), Ubuntu, Amazon Web Services (AWS), Google Cloud Platform (GCP), Docker, Fireblocks, AWS Lambda, Amazon EC2, WordPress, Hummingbot
Industry Expertise
Trading Systems, Project Management, High-frequency Trading (HFT), Binance Cryptocurrency Exchange, Marketing
Storage
Data Pipelines, PostgreSQL, InfluxDB, Google Cloud, Microsoft SQL Server, MySQL, Amazon S3 (AWS S3), Databases, Cloud Deployment
Other
Applied Mathematics, Mathematical Modeling, Linear Algebra, Partial Differential Equations, Analysis, Research, Data Science, Machine Learning, Finance, Algorithmic Trading, Mathematical Finance, Predictive Modeling, Time Series Analysis, Data Analysis, Reports, Trading, Stock Trading, Financial Modeling, Mathematics, Statistics, Data Analytics, Data Visualization, Artificial Intelligence (AI), Back-end, Programming, Dimensionality Reduction, Time Series, Data Mining, Statistical Modeling, Linear Regression, Quantitative Analysis, Quantitative Modeling, Quantitative Finance, Quantitative Risk Analysis, Regression Modeling, Backtesting Trading Strategies, Trade Finance, Statistical Analysis, Mathematical Analysis, Data, Data Modeling, Data Engineering, Regression, Classification, Data Scientist, Dashboards, Scientific Data Analysis, Algorithmic Trading Analysis, Statistical Data Analysis, Automated Trading Software, APIs, Cryptocurrency, API Integration, Forecasting, Fintech, Market Risk, Financial Markets, Risk Models, Monte Carlo Simulations, Web3, Smart Contracts, Decentralized Finance (DeFi), Analytics, Financial Analysis, Reporting, Simulations, Crypto, Technical Leadership, Leadership, Consulting, Mentorship & Coaching, Logistic Regression, Risk Management, Portfolio Analysis, Financial Software, Architecture, Time Series Data, Trading Bots, CEX, Ethereum Virtual Machine (EVM), Back-end Development, Liquidity Pools, Trade, Python, Trading, Quantitative Finance, Artificial Intelligence (AI), Feature Engineering, Modeling, Cryptocurrency Wallets, Uniswap, Options, Cryptocurrency Exchanges, Backtesting, Market Microstructures, Support Vector Machines (SVM), Gradient Boosting, Random Forests, Decision Trees, Google BigQuery, System Architecture, Code Review, Project Review, Parquet, Software Architecture, Agentic AI, RAG Systems, Large Language Models (LLMs), Vector Databases, AI Agent Orchestration, Pgvector, Agentic AI Systems, AI Agents, Metrics, Systems & Tools.Python, LLM Agents, Principal Component Analysis (PCA), Fourier Analysis, Stock Market, Technical Writing, Data Reporting, Clustering, Financial Data, Statistical Methods, Software Development, Cloud Services, Algorithms, Artificial Neural Networks (ANN), Heatmaps, Documentation, Technical Documentation, Writing & Editing, Bots, Financial Reporting, PDF Scraping, Scraping, Currency Exchange, Predictive Analytics, Stock Analysis, Hyperliquid, Real-time Processing, Low-latency Software, K-means Clustering, IT Management, FastAPI, AI Architecture, Facial Recognition, Credit Risk, Google Cloud Functions, Neural Networks, Linear Discriminant Analysis (LDA), WordPress Themes, Learning Management Systems (LMS), Elementor, Forms, WP Rocket, Astra, Vimeo, Front-end, Computer Vision, Deep Learning, WordPress Plugins, Education Management Systems, Deep Neural Networks (DNNs), Amazon Machine Learning, Cloud, Optimization, Linear Optimization, Web Scraping, Team Management, Product Management, Strategy, Telegram Bots, Cryptocurrency APIs, Derivatives, Futures, Futures & Options, Probability Theory, Forex Trading, Ethereum Smart Contracts, Liquid Staking Derivative (LSD), Quantitative Risk Modeling, WebSockets, Amazon Redshift, OpenAI Agents SDK
How to Work with Toptal
Toptal matches you directly with global industry experts from our network in hours—not weeks or months.
Share your needs
Choose your talent
Start your risk-free talent trial
Top talent is in high demand.
Start hiring