Yannick Pferr
Software Engineer @ Confluent
About Me
Welcome to my page!
My name is Yannick Pferr, I have a bachelors and masters degree in business information systems from TU Darmstadt and am currently working as a Full Stack Software Engineer in the ksqlDB team at Confluent.
I am passionate about software engineering and technology in general. In my freetime I like working on software projects and messing around with Raspberry Pi's or Arduino's. Other than tech, I really like working out, cooking and coming up with new healthy recipes (check out my food blog here).
If you want to get in touch, you can message me on LinkedIn, send me an email or simply use my contact form!
Looking for my resume? Look no further! Click the button below!
Experience
Confluent Inc.
since January 2023
Software Engineer | Full Time
- Full Stack Engineer ksqlDB
Commerzbank AG
October 2021 - December 2022
Software Engineer (IT-Trainee program) | Full Time
- Redesigned and rebuilt a supervised machine learning model for the treasury to predict if customers are going to use the special repayment option of their loan; boosted the F1-score of the old model by 22% (Python, scikit-learn, pandas)
- Developed and unit-tested the frontend of a foreign exchange trading app; improved maintainability and the code review process by enforcing coding style conventions (React, TypeScript, Styled Components)
- Implemented new REST API endpoints in the backend of the Commerzbank developer API portal and built a feature in the frontend that allowed users to see the deployment/staging status of their projects (React, Java Spring, REST APIs)
- Led a team of five interns/students to help different teams with the deployment of new APIs
Bosch Rexroth AG
January 2021 – June 2021
Data Scientist | Master Thesis
- Researched how machine learning can be used to optimize a production scheduling problem (combinatorial optimization); drafted, implemented, and evaluated a solution; completed the master thesis with the highest grade possible (1.0)
- Extracted machine and production data from one million excel files, analyzed, cleaned, and prepared the data for the model
- Implemented a reinforcement learning algorithm from scratch and achieved a 12% reduction in production time per production line which corresponds to a yearly decrease in costs of about 300.000 € per production line
- Analyzed results, translated them into business recommendations, and presented them to senior management
- Tech used: Python, JavaScript, pandas, TensorFlow, Keras, scikit-learn, Node-RED
Arlanis Reply AG
January 2018 – September 2021
Software Engineer | Bachelor Thesis, Intern, Working Student
- Built an app for consultants to help them automate various tasks (e.g.: data migration) in the Salesforce Marketing Cloud (SF MC) and saved them multiple hours of tedious manual work in the process (Java, SQL, REST/SOAP APIs)
- Developed a full stack web app for clients running in Marketing Cloud so clients could export specific data to excel files, without them needing to know how to write SQL queries (JavaScript in frontend & backend, HTML, CSS, SQL)
- Implemented a service that exported millions of campaign tracking records from SF MC, processed them, and imported them into SF CRM to help a client easily customize their campaign tracking reports (Java, Heroku, REST/SOAP APIs)
Skills
- Java
- JavaScript/TypeScript
- Python
- HTML/CSS
- SQL
- React
- Spring
- scikit-learn
- pandas
- TensorFlow
- Keras
- Next.js
- Gatsby.js
Projects
Commerzbank Collabothon: ReNetwork
Built a scalable web app in a team of five with Next.js, React and Firebase Firestore. The app connects refugees with volunteers through the in-app chat, to help the refugees with daily tasks that might be challenging for them. Won 1st place in the Google challenge with our idea "ReNetwork".
- JavaScript
- Next.js
- React
- Firebase
Deep Reinforcement Agent
Implemented two reinforcement learning algorithms (DDQN and REINFORCE) from scratch, one of which was used in my master thesis. The correctness of the implementation was tested with the test environments from OpenAI Gym.
- Python
- Machine Learning
- Reinforcement Learning
- OpenAI Gym
YPSwap: ERC-20 Token Exchange
Created my own ERC-20 token and deployed it on the Ethereum Ropsten Testnet. In the frontend you can exchange (test) Ether for YP coins, by connecting with MetaMask. The frontend was implemented with React and the token contract with solidity.
- Solidity
- JavaScript
- React
Blockchain System Explorer
Developed a Java app for researchers that can monitor & analyze the status of any blockchain platform using a client/server approach. The collected data can be viewed and queried through a web interface. The data is stored in InfluxDB and visualized with Influx Chronograf. System metrics of blockchain nodes can also be collected with the help of Influx Telegraf.
- Java
- InfluxDB
- Influx Chronograf
- Influx Telegraf
- Web3J
News Sentiment Analyzer
Built a web app with a team of students that searches for specific keywords in news articles, and then analyzes the sentiment of the text. The app uses Java, Apache Kafka and MongoDB to collect the data, Solr to analyze the text, and Redis and Node.js to visualize the data in a web interface.
- Java
- JavaScript
- MongoDB
- Apache Kafka
- Redis
- Solr
Magic Mirror
Built a smart mirror with a Raspberry Pi that could display various information. The backend was built upon a Node.js open-source library. Other than the basic features of the library, the mirror could play YouTube videos and Spotify tracks, and also be controlled through voice using Google STT.
- JavaScript
- Node.js
- Raspberry Pi
