Maksym Bevza, Developer in Kiev, Ukraine
Maksym is available for hire
Hire Maksym

Maksym Bevza

Verified Expert  in Engineering

Algorithms Developer

Location
Kiev, Ukraine
Toptal Member Since
November 30, 2021

Max has a Ph.D. and is a machine learning and NLP professional who excels with the back end, algorithms, and data structures. He has built numerous AI strategies for companies and implemented them with teams. His NLP experience includes GPT-3, PyTorch, Keras, and TensorFlow. Max has developed specialized products for Google, Facebook, and Grammarly. He enjoys devising reliable solutions to challenging NLP problems. Past roles have included Head of ML and Algorithms, CTO, CEO, and founder.

Portfolio

Knowt
Algorithms, Search Engines, AllenNLP, Amazon Web Services (AWS)...
WonderWay
JavaScript, Node.js, React, React Native, Next.js, Heroku, DevOps, SQL...
Develiry
Sales, Technical Hiring, Computer Vision...

Experience

Availability

Part-time

Preferred Environment

Ubuntu, Python, PyCharm, OS X

The most amazing...

...deep learning engine I've developed runs Grammarly's entire grammatical error correction function and replaced the old system.

Work Experience

Head of ML and Algorithms

2021 - PRESENT
Knowt
  • Collaborated with the CEO to build and implement the algorithm strategy of the company.
  • Built and entered into production many algorithm-heavy computations that involved NLP.
  • Optimized the entire processing pipeline 5-7 times to bring near-real-time processing.
  • Parallelized and distributed query processing to handle increased traffic.
  • Introduced agile processes to the algorithm team.
Technologies: Algorithms, Search Engines, AllenNLP, Amazon Web Services (AWS), Amazon S3 (AWS S3), Amazon SageMaker, DevOps, AWS DevOps, Python, Cython, C++, Stanford NLP, Generative Pre-trained Transformers (GPT), Natural Language Processing (NLP), SpaCy, NumPy, SciPy, Spark, Amazon DynamoDB, Parallel Programming, Artificial Intelligence (AI), Deep Learning

CTO and Co-founder

2020 - 2021
WonderWay
  • Developed an application that provides travel itineraries from bloggers and influencers (marketplace).
  • Built the MVP and pushed it into production in less than two months with no prior knowledge of mobile development.
  • Finished more than 40 customer interviews to detect and satisfy users' needs with the product.
Technologies: JavaScript, Node.js, React, React Native, Next.js, Heroku, DevOps, SQL, PostgreSQL, Artificial Intelligence (AI)

CEO and Founder

2019 - 2019
Develiry
  • Led the closure of seven deals for the development of machine learning solutions for SMEs and startups.
  • Iterated and tested the hypothesis of the value proposition to the level of acquiring users from the site.
  • Launched four developed MVPs to production that helped clients test their hypothesis of the technical feasibility and raise funding rounds.
Technologies: Sales, Technical Hiring, Computer Vision, Generative Pre-trained Transformers (GPT), Natural Language Processing (NLP), Python, C++, Chatbots, Object Detection, 3D Pose Estimation, Artificial Intelligence (AI), Deep Learning

COO and Co-founder

2019 - 2019
Shift AI
  • Led the closure of seven deals for the development of machine learning solutions for SMEs and startups.
  • Iterated and tested the hypothesis of the value proposition to the level of acquiring users from the site.
  • Launched four developed MVPs to production that helped clients test their hypothesis of the technical feasibility and raise funding rounds.
Technologies: Generative Pre-trained Transformers (GPT), Natural Language Processing (NLP), Technical Hiring, Sales, Sales Funnel, SpaCy, NumPy, Python, Artificial Intelligence (AI), Deep Learning

Research Scientists

2017 - 2019
Grammarly
  • Developed deep learning NLP systems that work on state-of-the-art levels for tasks Named entity recognition, error detection, and grammatical error correction using the Neural Machine Translation approach.
  • Built research infrastructure and tooling for the research team of 15 people.
  • Ensured successful delivery of the developed machine learning solutions to production.
Technologies: C++, Python, JavaScript, Bootstrap, Flask, Transformers, TensorFlow, PyTorch, SpaCy, NumPy, Pandas, AllenNLP, Docker, Amazon Web Services (AWS), Machine Translation, Neural Machine Translation, Artificial Intelligence (AI), Deep Learning

Software Engineering Intern

2016 - 2016
Google
  • Implemented unsupervised machine learning over Chrome extensions to cluster them and help detect malware in the Chrome Web Store.
  • Led cross-team communication to ensure the infrastructure team developed the necessary features based on the needs of our team.
  • Ensured the successful roll-out of the newly designed algorithm to production.
Technologies: C++, Python, D3.js, TensorFlow, Artificial Intelligence (AI), Deep Learning

Software Engineering Intern

2015 - 2015
Facebook
  • Developed the homepage dashboard, a place with charts containing key information for business managers such as total spend and total impressions for their ad account.
  • Collaborated with product managers to iterate on the product and ensure critical features were included.
  • Oversaw the roll-out of the new dashboard to production.
Technologies: JavaScript, React, HTML, CSS, PHP, SQL

Software Engineering Intern

2014 - 2014
Google
  • Modified the MapReduce framework, which our infrastructural team provided to other teams at Google.
  • Combined an arbitrary number of MapReduce into a single one improving the use of resources.
  • Created reports and presentations with the results of the optimization work to ensure smooth roll-out to all current recurring jobs.
Technologies: C++, MapReduce, Data Structures

Software Engineering Intern

2013 - 2013
Google
  • Developed a reference implementation of the Google Cloud Print device.
  • Implemented the end-to-end testing workflow for the newly developed device.
  • Created the help box for the Google Cloud reference implementation device.
Technologies: C++, Multicast DNS (mDNS), DNS Service Discovery (DNS-SD), HTTP, OAuth 2, XMPP

Neural Network 10x Speed Improvement Project

The neural network that performed an estimation of the entity's quality has been built using heavy BERT-based models that gave high accuracy but low speed. The goal was to optimize the processing with minimal sacrifices in its accuracy. To accomplish this goal, we gathered vast amounts of data and used the knowledge distillation to train a tree-based simple classifier to mimic the BERT-based model's predictions. It resulted in more than a 10x improvement in speed with less than one point of accuracy degradation, fully satisfying production requirements.
2017 - 2021

Ph.D. in Computer Science

Taras Shevchenko National University of Kyiv - Kyiv, Ukraine

2015 - 2017

Master's Degree in Computer Science

Taras Shevchenko National University of Kyiv - Kyiv, Ukraine

2011 - 2015

Bachelor's Degree in Computer Science

Taras Shevchenko National University of Kyiv - Kyiv, Ukraine

MARCH 2021 - PRESENT

Spacecraft Dynamics and Control Specialization by Colorado Bolder

Coursera

DECEMBER 2015 - PRESENT

Machine Learning by Stanford

Coursera

SEPTEMBER 2014 - PRESENT

Cryptography by Stanford

Coursera

Libraries/APIs

SciPy, TensorFlow, PyTorch, SpaCy, NumPy, Pandas, React, D3.js, Stanford NLP, Node.js, CatBoost

Tools

PyCharm, Amazon SageMaker

Languages

Python, C++, JavaScript, HTML, CSS, PHP, SQL

Storage

MySQL, Amazon S3 (AWS S3), Amazon DynamoDB, PostgreSQL, NoSQL

Platforms

AllenNLP, Ubuntu, Docker, Amazon Web Services (AWS), Heroku, OS X

Paradigms

MapReduce, DevOps, Parallel Programming

Frameworks

OAuth 2, Bootstrap, Flask, Spark, React Native, Next.js

Other

Software Development, Algorithms, Data Structures, Machine Learning, Software Engineering, Computer Science, Natural Language Processing (NLP), HTTP, Transformers, Artificial Intelligence (AI), Deep Learning, Generative Pre-trained Transformers (GPT), Cython, Multicast DNS (mDNS), DNS Service Discovery (DNS-SD), XMPP, Technical Hiring, Sales, Sales Funnel, Computer Vision, Chatbots, Object Detection, 3D Pose Estimation, Search Engines, AWS DevOps, Cryptography, Kinematics, Kinetics, Controlling, Machine Translation, Neural Machine Translation, BERT

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