Jakub B.

Educational content search engine and sharing platform

• Activities:

• Ideating and helping define overall startup strategy and UX
• Helping define OKRs
• Helping define database model and ontology structure
• Defining AI and application architecture, high-level algorithms
• Preparing for the switch to graph database and utilizing graph algorithms on top of Neo4j
• Productionization of research code
• Authority for GraphQL architecture
• Writing NLP and Graphs python and cypher code

• Algorithms:

• Tagger based on parts of speech and text rank
• Content difficulty level inference
• Language translator using mBART
• Recommender based on user activity embeddings scaled by activity type and amount
• Collaborative filtering based on user's skills, background, and activity
• Newsfeed based on content popularity, similar users' activity

• Technologies:

• Python, Transformers, PyTorch, Scikit-learn, Pandas, FAISS
• Neo4j, Graph data science library
• GraphQL, Graphene, Django, React, Apollo

Tool for researching startup investments and emerging technologies

• Techniques:

• Tagging startups by their description
• Extracting NER (money amounts, people, companies)
• Extracting sentences that provide the key information about the technological progress

• Technologies:

• Python, Spacy, PyTextRank
• PyTorch, Transformers, BERT

Projects for a non-profit organization developing civic service technology

• Project "Start teaching":

• Defining the overall project and UX strategy
• Defining the application and AI architecture
• Defining roles and hiring
• Planning and assigning technical tasks
• Code review
• Tech stack is Python, Django, React, GraphQL, Apollo

• Project "Innovations in journalistic storytelling"

• Advising on AI matters for all teams
• Leading brainstorming workshops
• Product ownership for project Audio news
• Devising technological roadmap
• Leading the recommendation engine efforts on top of Neo4j
• Presenting on behalf of the team
• Negotiating partnerships

• Project “Our politicians”

• Topic generation for news articles about politicians

Football analytics platform for scouting players. Advanced models as similar players, player types clustering, player ELO, expected goals, and many more

• Activities:

• Leading migration to Spark
• Developing state of the art football metrics
• Leading a team of full-stack developers and data scientists

• Technologies:

• FE in Angular 7
• BE in Python, SQLAlchemy, Flask
• Data transformation in PySpark, Parquet, DataBricks, Scala
• Models in Numpy, Scikit-learn, and Tensorflow
• Visualizations in D3.js

• Projects:

• Algo trading platform concept based on NLP analysis of financial reports and news articles
• Chess engine using alpha-beta pruned minimax written in Cython
• Own NER model by finetuning BERT
• Generating quiz answer possibilities using GPT-2
• Document similarity using BERT embeddings
• API creating n-gram “topics” from articles about politicians to provide overview of the news (project “Our politicians” by Cesko Digital)
• NLP stealth project analyzing sentiment analysis about particular qualities of a product
• App recognizing difficult foreign language vocabulary from book/text - hackathon project
• A platform for information extraction in Kotlin
• Chatbot for a virtual financial institution using GATE framework
• Chatbot intent recognition using FastText
• Cosine similarity-based recommendation in SQL
• Writing ETL SQL scripts, creating dashboards
• Refactoring current warehouse for 8 e-shops in E-Commerce Holding

• Frameworks

• FAISS, Solr, WordNet, FastText, GATE, Morphodita, Spacy
• Cheerio, Beautiful Soup, NLTK, TextBlob, Tensorflow, Keras
• Keboola, GoodData, Qlikview
• Kotlin

• Projects:

• React Native mobile app for NGO Rekonstrukce statu
• Strategy for organization prg.ai making Prague an AI superhub - case competition
• Sample selection from population algorithm for sociology research
• Typo correction tool for large questionnaire evaluation
• eGovernment project for Czech Ministry of the Interior in Java, MyBatis, Angular
• Frontend lead at key digitalization project in a traditional Czech bank
• Fintech startup website in React with server-side rendering using next.js
• Reverse engineering clipboard of Google slides to auto-generate a chart from data
• IT Strategy for the city of Prague prepared for municipal elections
• Mobile platform for discovering a city by local stories
• Digital agency website in Angular
• P2P lending project process analysis
• Browser plugin MindBrella for analyzing Facebook profiles, pages and groups for disinformation content
• Sales gamification mobile app in Ionic for parcel logistics company
• System for publishing public contract offerings for the state in Java

• Technologies

• Angular, React, Python, Java, Go, C++, Node.js, Swift, Express, Bootstrap, Flask, Django, Spring, Next.js
• React Native, Apollo, GraphQL, Ionic
• MongoDB, Firebase, MySQL, Sequelize, Mongoose, PyMongo, Neo4j

• Projects

• Defining product and UX strategy from big Czech retailer's online sales mobile app
• Football analytics platform for scouting talent
• Dating platform merging Reddit, Tinder, and Facebook
• Food ordering apps focused on sharing the delivery cost and optimizing delivery routes
• Discussion platform optimized for public policy issues
• City exploration app showing custom found stories about buildings and places

• Browser extension creating a foreign language textbook out of any text website. Personalized choice of sentences to be translated based on
your level. Automatically created vocabulary with the context where you saw the word. Complex vocabulary enrichment based on synonyms and topics.

• Technologies

• Angular, Python, Ionic
• Chrome and Firefox extension
• Recommendation engine
• Spacy
• Various Dictionaries and Corpuses

• Online scientific and political discussion/inquiry forum employing facebook level algorithms to order content according to quality, relevance, and interest.
• Organized by the inherent structure of wisdom - problems, solutions, and arguments. Plus facts are arranged in knowledge graphs.

• Technologies

• Angular, Apollo, Python, Flask, Graphene, Neo4j

• Technologies:

• Java and Javascript
• Angular, Node.js, Express, Spring, JSP, Bootstrap, JavaFX, Python
• MySQL, Sequelize
• Google Analytics
• Bot framework, QnAMaker, Table storage, Azure AD, GoodAI Chatbot, Graph API
• Ethereum Blockchain, Solidity, Truffle

• Projects:

• Codebook management system RDS in Java and Angular
• Preapproved loans lending system Impresto in Angular
• Blockchain demo project for connected car data management
• Several chatbots in Bot framework and Node.js, using MSFT Graph API