Peter Skvarenina, Developer in Frankfurt, Hesse, Germany
Peter is available for hire
Hire Peter

Peter Skvarenina

Verified Expert  in Engineering

Bio

Peter is a high-energy developer looking for exciting projects that he can help grow. He has experience working in both large corporations and startups, having worked for some of the best tech companies, like JetBrains, Nokia, and SUN, and contributed to many number-one solutions in their areas. After completing a master’s in computer science and machine learning, Peter is now expanding his knowledge of deep learning at Stanford, exploring its use to automate cognitive tasks.

Portfolio

Self-employed
PyTorch, TensorFlow, Keras, Amazon Web Services (AWS), Azure, Java, JavaScript...
JetBrains
Java, WebGL, Cloud, AI Programming, APIs, PostgreSQL, Image Processing...
Nokia
Java, Hadoop, Cloud, APIs, PostgreSQL, Data Pipelines, Image Processing...

Experience

  • Java - 12 years
  • Computer Vision - 8 years
  • Amazon Web Services (AWS) - 6 years
  • TensorFlow - 5 years
  • Deep Learning - 5 years
  • Machine Learning - 5 years
  • Natural Language Processing (NLP) - 4 years
  • Generative Pre-trained Transformers (GPT) - 3 years

Availability

Full-time

Preferred Environment

Visual Studio Code (VS Code)

The most amazing...

...project I've developed was Animatron.com, a JetBrains child company for which I designed and invented most of the algorithms.

Work Experience

Consulting VP of Engineering | Generative AI Consultant

2017 - PRESENT
Self-employed
  • Took part in a large-scale natural language processing (NLP) document search on AWS, replacing Elasticsearch, in collaboration with Amazon Kendra's team.
  • Contributed to an ML, cloud-based photo and vector editor (CV/NLP), whose functionality was demonstrated by NVidia at SIGGRAPH 2021, the premier conference and exhibition in computer graphics and interactive techniques (youtu.be/2fddvn16WIw?t=444s).
  • Consulted on a deep learning-based video editor with pose tracking–venture capital seed prototype.
  • Handled a microscopic manufacturing defect detection system that registered 97% accuracy in production.
  • Reconstructed 3D indoors using SLAM point cloud and semantic segmentation for an online room designer app.
  • Worked on a cloud-based fraud detection pipeline for mobile devices.
  • Contributed to developing contextual intelligent image content filters for ad providers.
  • Prepared a Telegram bot impersonating a licensed celebrity (GPT-4/LLaMA2, ElevenLabs TTS).
  • Developed a form-filling AI assistant with a talking avatar head guiding a user in filling in missing but required data (GPT-4, LangChain, GCP).
  • Listed among TIME Magazine's top 200 inventions of 2024. Researched and designed the LLM brain of the world's best LLM agents implementation—D-ID AI Agents, talking avatar head/upper body via RAG with hybrid search (GPT/Claude/LLaMA/Pinecone/ElevenLabs, etc.).
Technologies: PyTorch, TensorFlow, Keras, Amazon Web Services (AWS), Azure, Java, JavaScript, Natural Language Processing (NLP), Generative Pre-trained Transformers (GPT), Computer Vision, Large Language Models (LLMs), TypeScript, Artificial Intelligence (AI), C#, Image Generation, AI Programming, APIs, Pandas, Natural Language Understanding (NLU), Scikit-learn, Data Science, Data Pipelines, Image Processing, Generative Artificial Intelligence (GenAI), Hugging Face, Language Models, PDF, BERT, Generative Adversarial Networks (GANs), Stable Diffusion, Text to Image, Augmented Reality (AR), Artificial General Intelligence (AGI), OpenAI GPT-3 API, Generative Pre-trained Transformer 3 (GPT-3), Generative Pre-trained Transformer 4 (GPT-4), App Development, Full-stack Development, OpenCV, Fine-tuning, Data Scientist, Data Extraction, Data Engineering, System Design, ChatGPT, OpenAI GPT-4 API, Distributed Computing, CTO, Convolutional Neural Networks (CNNs), Image Analysis, Flask, Machine Learning Operations (MLOps), Architecture, Neural Networks, Data Loading, Full-stack, OpenAI API, Text to Task, LangChain, Chatbots, Amazon, Retrieval-augmented Generation (RAG), Speech Recognition

Principal Engineer

2012 - 2017
JetBrains
  • Designed the main algorithm of the Animatron animation software.
  • Produced the world’s first SVG SMIL editor–ranked #1 on HackerNews.
  • Presented a self-driving car technology at JetBrains' developer summit.
Technologies: Java, WebGL, Cloud, AI Programming, APIs, PostgreSQL, Image Processing, Web Development, App Development, Full-stack Development, OpenCV, Data Extraction, System Design, Distributed Computing, Image Analysis, Architecture, Neural Networks, Data Loading, Full-stack, Digitization

Senior SW Engineer

2008 - 2012
Nokia
  • Created a Google Earth-like 3D globe viewer for navigation purposes, including 3D cities, elevation, points of interest, and shortest path calculation.
  • Prepared the 3D elevation part of the Navigation Data Standard system–nds-association.org.
  • Co-designed the internal Nokia cloud, using Hadoop and Voldemort, while attending Navteq architecture summits.
  • Received the Special Achievement Award, 2010 edition.
Technologies: Java, Hadoop, Cloud, APIs, PostgreSQL, Data Pipelines, Image Processing, App Development, Full-stack Development, Data Extraction, System Design, Distributed Computing, Architecture, Scraping, Full-stack

Senior SW Engineer | Technical Lead

2007 - 2008
Sun Microsystems
  • Worked on a low-level clustering and transactional protocol of a JMS Grid, an enterprise messaging system used mostly in finance.
  • Saved over $100 million in damages by addressing customer issues in a time-critical manner.
  • Awarded for adoption, engagement with communities, and customer service in 2007.
Technologies: Java, Java Message Service (JMS), Distributed Systems, APIs, App Development, Full-stack Development, System Design, Distributed Computing, Architecture, Full-stack

Experience

Self-driving Car | Waypoint Navigation

https://github.com/EuROS-SDC17/Waypoint_Navigation
The capstone project for the Udacity's Self-driving Car Nanodegree that I attended in the first term of 2017. We used waypoint navigation in the simulator that was later run in an authentic 2016 Lincoln MKZ. I led an international team of five people, working on low-level car control with a robot operating system (ROS).

Demo available at:
www.youtube.com/watch?v=fQ-2ds-aInU

As-Rigid-As-Possible Shape Manipulation

https://github.com/squared9/Animation/tree/master/As_Rigid_As_Possible_Shape_Manipulation
Implementation of a fantastic algorithm for animating static characters (Igarashi 2005) from scratch in Python, using barycentric coordinates for movable handles. It should be combined with Computer Vision algorithms for background removal and shape outline detection for best results.

Semantic Segmentation for Road Detection

https://github.com/squared9/Self-driving-Car/tree/master/Semantic_Segmentation
Detection of road areas in images using Fully Convolutional Network (FCN)/UNet in Keras/TensorFlow, trained on the Kitty Road dataset. It is useful for self-driving cars and automated path-finding and navigation.

Adapt or Fail: A Tale of Three Transformer Adapters

A project from Facebook AI Research (FAIR, now Meta AI) about fine-tuning massive transformers using small pluggable adapters, leading to much faster fine-tuning times and the same or better performance.

Analyzing and Mitigating Dataset Artifacts

This is an NLP project from the University of Texas at Austin concerned with artifacts from spurious correlations and their removal using adversarial training sets. The project achieved significant improvement in disambiguating adversarial sentences from regular ones.

Education

2021 - 2022

Diploma in Artificial Intelligence

Stanford University - Stanford, CA, USA

2019 - 2022

Master's Degree in Business Administration (MBA)

University of Illinois Urbana-Champaign - Urbana-Champaign, IL, USA

2017 - 2020

Master's Degree in Computer Science

Georgia Institute of Technology - Atlanta, GA, USA

Skills

Libraries/APIs

Java Message Service (JMS), PyTorch, TensorFlow, Keras, Pandas, OpenCV, OpenAI API, Scikit-learn, WebGL

Tools

ChatGPT, LaTeX

Languages

Java, Python, JavaScript, C++, C#, TypeScript, R

Frameworks

Flask, Hadoop

Paradigms

Anomaly Detection, App Development, Distributed Computing

Platforms

Amazon Web Services (AWS), Amazon, Azure, Visual Studio Code (VS Code)

Storage

PostgreSQL, Data Pipelines

Other

Deep Learning, Machine Learning, Robotics, Cloud, Natural Language Processing (NLP), Computer Vision, Artificial Intelligence (AI), Large Language Models (LLMs), Optical Character Recognition (OCR), AI Programming, APIs, Image Processing, Generative Artificial Intelligence (GenAI), Chatbots, Web Development, Linear Regression, Language Models, BERT, Generative Adversarial Networks (GANs), Chatbot Conversation Design, Text to Image, Generative Pre-trained Transformers (GPT), Artificial General Intelligence (AGI), OpenAI GPT-3 API, Generative Pre-trained Transformer 3 (GPT-3), Full-stack Development, Fine-tuning, Data Scientist, System Design, Image Recognition, Text Recognition, Convolutional Neural Networks (CNNs), Image Analysis, Architecture, Neural Networks, Facial Recognition, Data Loading, Variational Autoencoders, Deep Neural Networks (DNNs), Full-stack, Digitization, Analytics, Text to Task, LangChain, Retrieval-augmented Generation (RAG), Innovation, Stable Diffusion, Natural Language Understanding (NLU), Data Science, Reinforcement Learning, Financial Forecasting, Image Generation, Hugging Face, Generative Design, Augmented Reality (AR), Object Detection, Generative Pre-trained Transformer 4 (GPT-4), Data Extraction, Data Engineering, OpenAI GPT-4 API, CTO, Machine Learning Operations (MLOps), Speech Recognition, Entrepreneurship, Leadership, Finance, Distributed Systems, Robot Operating System (ROS), Natural Language Queries, Transformers, GPU Computing, PDF, Text to Video, Scraping

Collaboration That Works

How to Work with Toptal

Toptal matches you directly with global industry experts from our network in hours—not weeks or months.

1

Share your needs

Discuss your requirements and refine your scope in a call with a Toptal domain expert.
2

Choose your talent

Get a short list of expertly matched talent within 24 hours to review, interview, and choose from.
3

Start your risk-free talent trial

Work with your chosen talent on a trial basis for up to two weeks. Pay only if you decide to hire them.

Top talent is in high demand.

Start hiring