Inline Manual

Inline Manual provides a service to guide, support and engage prospects and customers right within an application. Easily create product walkthroughs, on-boarding guides, new features announcements, tooltips and manageable product documentation.

Used and loved by small and big companies to get customers up to speed in no time, available for any web based product. All this without any coding.

Some of the companies using Inline Manual: Pearson, pipedrive, eToro, Miovision, University of Michigan, iForex and many others from all around the world.

Founder

Eventhunt.io

Eventhunt is a Product Hunt like platform for tech/startup events in your city. It is for people who want to socialize, learn new things, get new opportunities (investors, free business advise) in the area they are in and also to help to get to know people when they move to a new location or want to get more involved while traveling.

Co-founder

Frontendunited

Organiser - travelling conference for designers, developers & drupalistas from all kinds of backgrounds - focused workshops, talks & sessions on design, frontend engineering & Drupal. (2010 Prague, 2011 Berlin, 2012 Amsterdam, 2013 UK)

Organiser

Atomic Ant Limited

Drupal developer, designer, themer, project manager

Owner

UBS Wealth Management

While being a senior Drupal developer, I started an initiative to bring into our workflow user centric design.

Senior developer, pushing User Centric Design

Drupal development, improving user experience/usability of an internal support portal. Improving processes when it comes to deployment and analysis.

Senior web developer

Drupal.org

Developer

Turner - Capgemini

Front-end and Drupal development for HLNTV.com

HLN TV

BrightTALK

Drupal developer

freelancer

Drupal developer, web designer

I'm the Founder of Inline Manual. I've worked directly with hundreds of clients serving millions of users each year. My experience working in front-end development and design lead me to user experience. Soon I realized what was missing: a way to allow the marketing and content teams to create in-app guidance on the fly. With Inline Manual, we help our clients improve onboarding and feature adoption, in a flexible and customizable way. Message me about how I can implement a better user-onboarding for your end-users with Inline Manual.

Marek Sotak

Fritz The AI Guy

- For Upwork: While in Marketing team, implemented a new version of the CLTV (Customer Lifetime Value) model. While in discovery team, improved job post’s recommended freelancers models.
- For Portal of Public Purchases: The Portal of Public Purchases is
a Brazilian GovTech that supports a platform with the same name
that allows services/products’ suppliers to bid for public
contracts. We started working together in January 2022, when they
came up with a business problem: The currently implemented
recommendation system (which recommends contracts to qualified
suppliers) is terrible. Since then, I worked in a Proof-of-Value
to make a better recommendation system using suppliers' previous
bids and contract’s item descriptions to assess qualification
through description similarity. Currently, the project is
entering the MVP phase;

- For BIUD Marketing: Development of an API to segment clients using RFM clustering from brazilian invoice data. This API was developed using BentoML;

- For Bendix: Performed segmentation of real estate clients based on their purchase behaviour and demographics. The clusters were identified using KMeans and Silhouette Score optimization on their purchase features. To explain the clusters based on demographics, a XGBoost model was trained using the demographics as features and the clusters (previously identified) as targets. A SHAP TreeExplainer was applied to the trained model, in order to identify which demographics had the most impact on the model predictions.

Founder & ML/AI Engineer

iNuTech - NuTech Institute of Applied Research in Science, Technology and Innovation

- Development of ML pipelines to train and deploy a SOTA deep learning model to identify decisions, jurisprudence and legislations on brazilian legal documents;

- Building APIs to integrate the ML pipelines with Atrium: A system to help lawyers on the writing of legal documents with bigger chances of winning on court;

- Participating on the framing of data science objectives for projects on IoT and personalized education.

Machine Learning Researcher

YData

- Lead the data team by providing mentoring to colleagues, defining
team allocation, interfacing with related teams (e.g.,
engineering), interfacing with clients (e.g., business problem
assessment, results presentation), defining the library
architecture, managing research projects, managing
proof-of-values, doing technical assessments for new candidates;

- Development of proof-of-values for clients interested in
synthetic data generation for tabular data;
- Implementation of new functionalities (from new papers, ideas,
brainstorms, or client’s feedbacks) in the internal library.
Also, interfacing with the engineering team to guarantee that the
updates run smoothly on YData’s platform;
- Research on time series data synthesization: I developed a novel
method to generate multivariate, multi entity time series using
Dynamic Bayes Nets.

Senior Machine Learning Engineer

Dephion

Dephion is a startup focused on developing elegant digital solutions to advance mental and physical well-being, mostly related to work well-being. During the time I worked as a Data Science Specialist (in a freelance contract) for them, I developed an activity recommender system for their app. The context of this development was the following:

1. The app had activities on it, which the user could interact with;
2. As the app was still in development, we didn't have user data at all;
3. All the activities on the app were based on research articles, which I had at disposal.

With this in mind, I created a similarity network (where activities were nodes and similarities were weighted edges) based on the research papers associated with each activity and segmented the network with a graph-based clustering algorithm.

Based on the network, the clusters and custom score metrics, we were able to recommend activities in three ways:

1. Reinforce: Which reinforced what the user would previously have seen;
2. Related: Which gave activities that was different from recent activity, but in the same cluster;
3. New: Which gave totally new activities to the user, in opposition to what it was doing before.

During cold start, the user would be recommended activities in a rule-based manner. As soon as the user achieved a certain number of activities in its history, the recommender system would start operating.

Besides doing the experimentation, I also put the system in production on Azure environment following the Company's software engineering best practices and necessities.

Machine Learning Specialist

Kunumi

- Promoted from Data Scientist to lead 4 data scientists and 1 AI Strategist;

- Development of a complex time-series ML system to generate, for Points of Sale, daily order recommendations for products based on demand prediction in order to avoid missing products on the store’s stock;

- Development of ML projects on the retail sector in the order of 4mi BRL (~1mi USD), generating business answers, automating analysis and, developing ML models;

- BeHy speaker: BeHy is a course directed to stakeholders (C-levels, directors, VP’s) with the purpose to technically educate them about AI and its impact in today’s world. As a speaker, I’m responsible to prepare the technical part of the course material and present them to stakeholders in an intuitive way, so they can grasp the fundamental concepts about AI;

- Co-development of business plans and pitch decks for new AI Ventures (most of them related to retail);

- Participation on the technical cooperation team, responsible to present and discuss business cases, and to develop new management tools to help in key stages of data science projects life cycle.

Tech Lead

- Worked in the early stages of Kunumi’s COVID-19 death toll prediction platform. The team was composed by researchers, data scientists, ML engineers and domain specialists. My role in this project was, having the validated researchers prototypes at hand, to structure the production pipelines for feature engineering, modelling and results reporting. Besides that, I contributed to the project with suggestions on how to model the problem, so we would have the best interpretability of the results. The pipelines were structured using Kedro. The most recent version of this project can be found here: www.covid-19.kunumi.com;

- Development and deployment of data and ML pipelines in environments like Databricks and Kubernetes;

- Development of a Python library to automate the process of clustering arbitrary datasets, while reaching the best clustering results (in terms of interpretation), and reporting results;

- Basket analysis using complex networks to answer questions like: Which products are the most important?, Which products connect different groups of clients?, What are the products that most explain some segment of clients?, Which products could be sold together in order to leverage sales?

Centro de Gestão e Estudos Estratégicos

- Worked on the Sustainable Cities program in partnership with the UN. As the data scientist of the team, I was responsible to frame, from the business problems, the data science objectives of the project. Basically, the project had two main data science objectives: To discover sustainable innovation faster and to understand how Brazil is divided in terms of natural resources, economic metrics and public policies;

- Development and maintenance of a microservice using Docker, RabbitMQ and the Elastic Stack. This service was responsible to track tweets about sustainable innovation in real time and to identify the degree of innovation of the tweets;

- Development of a clustering prototype capable of segmenting Brazil based on natural resources and economic metrics. The prototype was created using KMeans and Silhouette score optimization the find clusters. It was also possible to define, using parameters, how much a specific feature should impact in the final clustering result.

Data Scientist

- Used network science and scientometrics to find relevant researchers and research articles in different knowledge areas. Moreover, used causal impact to identify if public financing of research projects, in fact, brought return in terms of research advancements.

- Creation of several prototypes to analyse research papers using complex networks and natural language processing;

- Dashboard design and development for the internal systems;

- Development of unstructured data processing pipelines;

- Development of a web crawler to download paper metadata from Web of Science.

Data Science Intern

Legal Labs Brasil

- Design and Development of the early stage of a big data pipeline to apply OCR in 6 TB of PDFs. Those PDFs were used to train deep learning models to classify legal documents;

- Development and maintenance of several web crawlers to download legal documents from regional court’s websites. The websites were completely different from each other and each of them required a specific web crawler. The web crawlers were developed on Node.js and were integrated to the internal crawling system, which was based on an OpenFaaS platform;

- Development of a tagging system using Vue.js, Node.js and GraphQL. This system was used internally to allow lawyers to tag documents with ease and in a structured way;

- Review of Dra. Luzia’s, an AI system to facilitate the work of lawyers, core deep learning model, responsible to classify legal documents, in order to give a perspective on the correctness of the modelling process.

Machine Learning Engineer

I am a 26 years old Tech Lead from Brazil: Working with data since I was 18 years old, I have accumulated an outstanding amount of professional experience for someone my age, taking on leadership roles in companies and projects I was involved with. Over the years I had the opportunity to develop projects across diverse domains like scientometrics, law, sustainability, health, and more recently retail.

In my work I aim to combine my theoretical understanding of multiple Artificial Intelligence and Machine Learning techniques with the practical implementation details that are required to build excellent products that add value to my stakeholders. In my experiences over the years, I have worked with a great number of tools and techniques: From complex networks to deep learning, from PostgreSQL to Cassandra, from Docker to Databricks.

My goal is to further develop my leadership skills and hone my expertise in MLOps, solving business challenges with scalable and performable ML pipelines, a field I have already started exploring both as a self-learner and as a consultant trying to adapt to the demands of my data products. Progress is what makes me wake up everyday and give my best!

taplio

The Battle

Marek Sotak

Gabriel Fritz Cavalcante

Followers

Posts

Engagements

Likes

Ready to grow your LinkedIn presence?