Maksim Tsvetovat, Artificial Intelligence (AI) Developer in Washington, DC, United States
Maksim Tsvetovat

Artificial Intelligence (AI) Developer in Washington, DC, United States

Member since May 18, 2021
For over 25 years, Max has worked on the cutting edge of computer science, machine learning, and modeling and simulation. He has delivered AI and ML tools to clients from hospitals to the federal government and hedge funds. Max looks beyond the hype and buzzwords with real-world experience and applies that experience to his projects. Max is both a developer and a leader—he enjoys working with a team and writing code. He's looking forward to bigger challenges and learning opportunities.
Maksim is now available for hire




Washington, DC, United States



Preferred Environment

Python, Keras, LSTM Networks, Natural Language Processing (NLP), MongoDB, PostgreSQL, Linux, BERT, ImageNet, NumPy

The most amazing...

...tool I've built is an AI-driven game that teaches disabled kids to exercise within their abilities and limitations. The kids' smiles are priceless!


  • Associate Professor (part-time)

    2019 - PRESENT
    George Washington University
    • Designed and taught a 2-course sequence on programming for analytics and machine learning and a course on simulation modeling for systems thinking.
    • Ensured my students, who come from diverse backgrounds and little to no programming experience, leave with a comprehension of data wrangling, classic analytics methods, and statistical machine learning through my Programming for Analytics course.
    • Taught the principles of neural networks and their applications in NLP, computer vision, time-series analysis, and algorithmic trading in this "advanced beginner" course, Neural Networks and Deep Learning.
    • Taught various modeling methods from systems dynamics and differential equation methods to microsimulation, agent-based modeling, and large-scale mixed-method simulation models in the advanced course, Simulation Modeling.
    Technologies: Machine Learning, Artificial Intelligence (AI), Neural Networks, Data Analysis, Keras, Simulations, Agent-based Modeling, Python, NumPy, TensorFlow, Pandas, SciPy, Plotly, NLTK, Scikit-learn, Regex, System Architecture, Text Categorization, Behavioral Science, Data Science, Databases, Signal Processing, PyTorch, Models, Image Processing, signal video analysis
  • CTO/Chief Scientist

    2013 - PRESENT
    Open Health Network
    • Led the creation and deployment of a system for practitioners to upload videos to a secure server, analyze content with a neural network, and extract relevant data. The major medical center analyzes >5,000 hours of video psychotherapy sessions.
    • Created a first-in-its class conversational interface (chatbot) that administers a 12-session motivational interviewing and cognitive behavioral therapy program to assist patients in a smoking cessation program.
    • Conducted extensive analysis of randomized clinical trials, analyzed the data using neural network text mining, and summarized using natural language generation to present to the client for a pharmaceutical co. requiring an analysis of its strategy.
    • Designed a set of algorithms analyzing cross-provider linkage patterns in claims that correspond to fraudulent transactions in opioid prescriptions and durable health equipment, estimate the risk of fraud, and refer cases to investigators.
    • Supervised a team of five developers and two data scientists. Ran an Agile and Kanban workplace with a distributed team in four different time zones.
    Technologies: Python, Neural Networks, Deep Learning, Healthcare IT, Wearables, Sensor Networks, Keras, MongoDB, Linux, NumPy, TensorFlow, Pandas, SciPy, Plotly, NLTK, Scikit-learn, System Architecture, Text Categorization, JavaScript, Behavioral Science, Data Science, Architecture, Databases, REST, AWS, Signal Processing, Video Processing, Health, Models, Mobile
  • Data Science Consultant

    2015 - 2016
    Global Consumer Packaged Goods Company (under NDA)
    • Served as a part of a 3-member data science team; analyzed CPG supply and demand data across the Latin American market. Utilized hyper-local data on weather, holidays, sports events and traffic to build comprehensive demand forecasting tool.
    • Created a detailed map of product needs, demand, complementarity, and substitution using geospatial and time-series analysis. The resulting tool was used to pre-position supply trucks to optimize product sell-through and minimize retailer inventory.
    • Analyzed the impact of signage and other promotional programs, as well as spill-over effects on competitive products.
    • Developed and deployed the system that the client then spun off into a separate company that Salesforce has acquired.
    Technologies: Consumer Packaged Goods (CPG), Demand Forecasting, Artificial Intelligence (AI), Recurrent Neural Networks, Time Series Analysis, Geospatial Data, Geospatial Analytics, Keras, PostgreSQL, Linux, Python, NumPy, TensorFlow, Pandas, SciPy, Plotly, NLTK, Scikit-learn, System Architecture, Text Categorization, Data Science, Architecture, Databases, REST, AWS, Models
  • Text Analysis and NLP Data Scientist

    2010 - 2013
    Real Capital Analytics
    • Took over an antiquated REGEX-based news analysis and filtering system; revamped using emerging ML-based text analysis and classification tools. Extracted structured data from unstructured text.
    • Reduced the time to analyze, process, and enter into the DB a data item from 5 min to 30 seconds with the resulting tool. Reduced the amount of rejected and irrelevant data from 60% in the legacy system to 24% in ML-based systems.
    • Processed news items automatically linked to GIS, resulting in a real-time understanding of newsworthy developments in a specific geographical area.
    • Reduced labor costs in the information processing team by approx. $4 million in the first year of system deployment while reducing the data processing backlog from about one week to zero.
    • Built ML capabilities the client used as key IP in their merger, creating approx $10-15 million of additional value to investors.
    Technologies: Python, Scikit-learn, TensorFlow, SpaCy, NLTK, Natural Language Processing (NLP), Text Classification, Data Extraction, GIS, Keras, MongoDB, PostgreSQL, Linux, Agent-based Modeling, NumPy, Pandas, SciPy, Regex, System Architecture, Text Categorization, JavaScript, Data Science, Architecture, Databases, REST, AWS
  • Assistant Professor/Computational and Data Science

    2005 - 2012
    George Mason University
    • Developed five courses for my department, including Complex Systems 101, Social Network Analysis, Complex System Analysis and Modeling, Modeling of Human Behavior and Organizations, and Machine Learning for Simulation Modeling.
    • Authored a textbook on social network analysis; topped at #48 on Amazon's data science best-seller list.
    • Oversaw graduate school admissions process for interdisciplinary doctoral, master's, and certificate programs.
    • Conducted research in the use of simulation modeling to understand the behavior of complex organizations and social systems. Published over 50 peer-reviewed papers on various topics in the area.
    • Supervised a group of graduate students on multiple NSF and DARPA-funded research efforts.
    Technologies: Simulation modeling, Machine Learning, Artificial Intelligence (AI), Simulations, Agent-based Modeling, Python, NumPy, System Architecture, Text Categorization, Behavioral Science, Data Science, Databases


  • Advanced Conversational Interfaces

    Cognitive-behavioral therapy (CBT) and motivational interviewing are powerful techniques for addressing mental and physical health challenges such as smoking cessation, obesity, and eating disorders. However, in-person CBT is expensive and not accessible to many patients.

    OHN has created a first-in-its class conversational interface (chatbot) that administers a 12-session motivational interviewing and CBT program to assist patients in a smoking cessation program. TAMI is undergoing a randomized clinical trial in 2021.

  • Remote Exercise Testing

    I worked with a major research hospital to design a set of tools to approximate the level of cardiovascular health (VO2Max) remotely, utilizing commodity exercise bikes, sensors, and mobile phones.

    The process uses a sequence of specific exercise actions (pedal speed, gear, etc.) and heart rate and takes approx. 10 minutes to complete. The results are processed with a custom-trained machine learning model to approximate VO2Max without the use of site visits or expensive gas exchange sensors.

    The process is currently in clinical testing, estimating at 90% accurate compared to gas exchange tests.

  • Advanced Image and Video Analysis

    A pediatrics department major academic medical center manually analyzes more than 5,000 hours of video recorded psychotherapy sessions. With a growing workload, the emergence of telehealth, and the need to support remote interventions during COVID-19, an streamlined solution was required
    OHN has created and deployed a system that allows practitioners to upload videos to a secure server. Videos are tagged and transcribed automatically utilizing neural networks. A licensed practitioner reviews and approves the results; time to analyze a session and prepare a report has been reduced from 24 minutes to six minutes. OHN will be maintaining and expanding the system to a larger number of practitioners.

  • Natural Language Understanding

    We have created a chatbot that uses a set of open-ended questions presented to 1,800 US government employees across 60 field offices in four US regions. The responses were mined using neural network topic modeling. They were collated into 11 major topic areas regarding adopting electronic medical records and digital health technologies, resulting in strong policy recommendations made to senior leadership.

    We created the chatbot and created the neural network topic modeling engine utilized in this project. We are maintaining the system and supervising its deployment to additional U.S. Navy and Air Force customers.


  • Languages

    Python, Regex, Python 3, JavaScript
  • Libraries/APIs

    Keras, NumPy, TensorFlow, Pandas, SciPy, Scikit-learn, SpaCy, NLTK, PyTorch
  • Tools

    Plotly, GitLab CI/CD, GIS
  • Paradigms

    Agent-based Modeling, Data Science, REST
  • Platforms

    Amazon Web Services (AWS), Mobile, Linux
  • Storage

    Databases, Data Pipelines, MongoDB, PostgreSQL
  • Industry Expertise

    Project Management
  • Other

    LSTM Networks, Natural Language Processing (NLP), BERT, Artificial Intelligence (AI), Neural Networks, Simulations, Heuristics, Deep Learning, Healthcare IT, Regression Modeling, Computer Vision, Convolutional Neural Networks, Transformers, Chatbots, Machine Learning, Data Analysis, Simulation modeling, Text Classification, Text Categorization, System Architecture, Data Extraction, Recurrent Neural Networks, Time Series Analysis, Behavioral Science, Architecture, Big Data, Leadership, Security Clearance, AI Design, Health, Models, ImageNet, Wearables, Sensor Networks, Conversational Interfaces, Cognitive Behavioral Therapy (CBT), Video Analysis, Consumer Packaged Goods (CPG), Demand Forecasting, Geospatial Data, Geospatial Analytics, AWS, Data Engineering, Signal Processing, Video Processing, Image Processing, signal video analysis, Bluetooth
  • Frameworks



  • Ph.D. in Computer Science
    1999 - 2005
    Carnegie Mellon University - Pittsburgh, PA

To view more profiles

Join Toptal
Share it with others