Alexander is available for hire
Hire AlexanderAlexander Lichtenstein
Verified Expert in Engineering
Data Scientist and AI Developer
Frankfurt, Hesse, Germany
Toptal member since October 4, 2021
Bio
Alex is a research scientist, applied AI engineer, and technology leader with 20+ years in software engineering. With an academic background in computer vision, his expertise spans mathematical modeling, optimization, predictive analytics, NLP, transformers, and training and fine-tuning vision and language models. He translates business challenges into practical AI products, shapes product direction, and leads teams from research and experimentation through evaluation and delivery.
Portfolio
Under NDA
AI Agents, Ontologies, Model distillation, LoRa, LangChain...
Under NDA
Machine Learning, Python, Machine Learning Automation...
Lg Nova
Artificial Intelligence (AI), DeepSeek...
Experience
- Machine Learning - 10 years
- Python - 10 years
- Neural Networks - 5 years
- Generative Pre-trained Transformers (GPT) - 5 years
- Amazon Web Services (AWS) - 4 years
- ChatGPT - 2 years
- LangChain - 2 years
- DeepSeek - 1 year
Work Experience
Senior AI Engineer (Contract)
2025 - 2026
Under NDA
- Developed agentic AI workflows for healthcare applications, coordinating AI agents across multi-step tasks.
- Contributed to model distillation for healthcare AI, transferring capabilities from larger models into smaller, more compact models.
- Designed knowledge graph architectures to structure medical information and represent relationships between healthcare concepts.
- Implemented a software engine for a healthcare AI project, translating domain requirements into application logic and processing workflows.
Technologies: AI Agents, Ontologies, Model distillation, LoRa, LangChain, Large Language Models (LLMs), Large Language Model Operations (LLMOps)
Python | ML developer to build a capacity planning model
2023 - 2026
Under NDA
- Designed and delivered a production-scheduling web application for a manufacturing plant, automatically recommending the sequence and resource allocation for every production batch.
- Built an optimization engine that converts the full customer order book into a feasible production plan in under a minute, accounting for changeover rules, equipment capacity, delivery dates, and existing inventory.
- Developed an interactive planning interface that lets planners adjust the recommended plan and see the daily schedule and Excel reports update instantly.
Technologies: Machine Learning, Python, Machine Learning Automation, Supervised Machine Learning, Capacity Planning, Integer Programming, Linear Optimization, Dynamic Programming
GenAI DeepSeek Expert
2025 - 2025
Lg Nova
- Built a GenAI MVP that creates personalized employee training courses.
- Automated interviews, knowledge assessment, and course creation with a multi-agent pipeline.
- Integrated fast LLM inference, document extraction, and interactive dashboards.
Technologies: Artificial Intelligence (AI), DeepSeek, Generative Artificial Intelligence (GenAI), Datasets, OpenAI, Meta Llama, Mistral AI, Groq, Tesseract, Dash, Flask
ML Developer (GenAI)
2024 - 2025
Under NDA
- Built a GenAI chatbot for querying real-time IoT sensor data.
- Connected Amazon Bedrock to OpenSearch for vector retrieval and Redshift for SQL queries.
- Refined prompts for accurate, domain-specific responses.
- Provisioned scalable infrastructure with AWS CloudFormation.
Technologies: Amazon Web Services (AWS), Amazon Bedrock AgentCore, Redshift, Amazon OpenSearch, AWS CloudFormation, Python, SQL
Computer Vision Developer
2023 - 2024
Agnico Eagle Mines Limited
- Developed a YOLO-based wildlife detection model for environmental reporting.
- Trained a custom filtering model to reduce false positives.
- Set up a Databricks processing pipeline to process a large amount of data in batches.
Technologies: Computer Vision, Python, Databricks, AI Adoption
Lead AI Developer
2023 - 2023
Under NDA
- Co-developed a platform for automation based on LLM agents. It allows users to create an agent, fill it with knowledge, and connect it to actions, e.g., sending emails, checking CRM, etc.
- Led the back-end team and developed the architecture of the solution together with the CTO.
- Fine-tuned basic models for cost reduction and speed.
Technologies: Open-source LLMs, Large Language Models (LLMs), Large Language Model Operations (LLMOps), OpenAI GPT-4 API, OpenAI, ChatGPT, Python 3, FastAPI, RabbitMQ, AI Agents, AI Adoption, Strategy
Machine Learning Expert
2022 - 2023
Health Access LLC
- Trained a CNN model to detect early eye problems using red-eye-reflex pupil images.
- Created a back end for the app based on Azure and then moved it to the AWS cloud (per client request).
- Provided fixes to the existing database architecture.
Technologies: Machine Learning, Deep Learning, Image Analysis, Python, Azure API Management, Azure, APIs
Data Scientist
2021 - 2023
Industrial Machinery and Bearing Production Company (Under NDA)
- Developed a machine learning proof of concept to predict large customer orders and support demand forecasting and resource planning.
- Developed a predictive maintenance model using machinery signal analysis to identify early signs of equipment degradation and support proactive maintenance planning.
- Implemented a proof-of-concept model at SKF using genetic algorithms and dynamic programming to optimize employee travel routes within the plant, targeting reduced travel distances and improved operational efficiency.
Technologies: Time Series Analysis, Signal Processing, XGBoost, Dynamic Programming, Data Science, NumPy, Pandas, Databases, Jupyter Notebook, PyCharm, Artificial Intelligence (AI), AI Adoption, Linear Optimization
CTO | Team Lead
2016 - 2021
Aogency
- Created MVPs for eight startups. Some of these MVPs have been based on ML solutions. Implemented these solutions using such technics as deep learning, regression, neural networks, and natural language processing.
- Built and managed a team of 5 to 10 developers and oversaw the whole work process; mastered my skills in different Agile methodologies.
- Implemented multiple software solutions for big corporate clients.
Technologies: Team Leadership, Technical Leadership, Python, Machine Learning, Microservices, Flask, Google Cloud, Amazon Web Services (AWS), NoSQL, Docker, SQL, PyTorch, Databases, Jupyter Notebook, PyCharm, AI Adoption, Strategy
Data Scientist | Software Developer
2020 - 2020
Data Science Startup (Under NDA)
- Implemented automated data collection and aggregation from open sources, including both professional and academic ones.
- Created an ML model (NLP) for data extraction from unstructured CVs.
- Implemented data ML pipelines for processing and validation of the data coming from different sources.
- Co-developed the platform architecture, database structure, and API.
Technologies: Python, Generative Pre-trained Transformers (GPT), Natural Language Processing (NLP), Web Scraping, Blazegraph, Databases, PyCharm, Artificial Intelligence (AI), AI Adoption
Consultant
2018 - 2018
Consulting Company, Nature Sciences in Finance (under NDA)
- Provided consulting on IT transformation to the client.
- Acted as a part of a knowledge team and supported communication to the IT experts team.
- Created stepwise change requests for the system transformation to the IT department.
Technologies: Scrum, Concepting, AI Adoption, Strategy
Scientific Employee
2013 - 2017
Frankfurt Institute for Advanced Studies
- Developed a biomechanical model of human eye muscles with a rendering component; simulated different disorders of the visual system and the development of the visual cortex.
- Gave multiple talks at different conferences and published papers.
- Showed that the application of the principle "less is more" brings advantages in learning neural networks.
Technologies: MATLAB, Python, C++, OpenGL, Machine Learning, Reinforcement Learning, Neural Networks, Modeling, Computer Vision, Science
Technical Lead
2008 - 2012
Stimultest
- Developed an algorithm that measures personal attitudes toward various visual stimuli.
- Created an IT team and built an MVP of the project.
- Provided full cycle project deployment and support.
Technologies: C++, Qt, Statistics
Project History
Real Estate Platform
Developed a platform for the client from the real estate industry, including the back end, admin panel, and app. The main purpose of this app was to connect tenants and landlords, as well as to give a price estimation for real estate and make forecasts about the price development in the following years. Also, the platform was able to perform automatic data extraction from tenant contracts and utility bills (as photos) and add this information to the database on the fly.
RESPONSIBILITIES
• Regression model for price estimation and forecasting
• NLP model for data extraction from the documents
• Programming of the front-end part (Ionic app)
• Programming of the back end
• Testing and deployment of the project
RESPONSIBILITIES
• Regression model for price estimation and forecasting
• NLP model for data extraction from the documents
• Programming of the front-end part (Ionic app)
• Programming of the back end
• Testing and deployment of the project
Mobile App for Face Verification
Developed a mobile app for disability card verification. This app's primary function is cross-checking a person's face and disability cards. After successful verification, a personal ID code was generated.
The main technical issue in this app is comparing selfies and disability card photos due to the different resolutions and quality. The problem was solved with the help of deep learning.
RESPONSIBILITIES
• Team lead.
• Programming of microservices and deployment on the cloud (Flask)
• Implementation of the deep learning model for the comparison of photos
• Testing and deployment of the project
The main technical issue in this app is comparing selfies and disability card photos due to the different resolutions and quality. The problem was solved with the help of deep learning.
RESPONSIBILITIES
• Team lead.
• Programming of microservices and deployment on the cloud (Flask)
• Implementation of the deep learning model for the comparison of photos
• Testing and deployment of the project
Mobile App (Big Corporate)
Development of the mobile app for a big corporate client. The main purpose of this app is to provide content on health topics to the end users. Our implementation was based on Firebase (Google Cloud). In addition, we have integrated the customer database to provide access to existing clients.
The content of the app is controllable through the admin panel.
Responsibilities:
• Project architecture.
• Team lead.
• Programming microservices and deploying on the cloud.
• Implementation of the recommendation system (based on articles tags, ML solution) for the end user.
-Testing and deployment of the project.
The technology stack consisted of Firebase, Flutter, and GCP.
The content of the app is controllable through the admin panel.
Responsibilities:
• Project architecture.
• Team lead.
• Programming microservices and deploying on the cloud.
• Implementation of the recommendation system (based on articles tags, ML solution) for the end user.
-Testing and deployment of the project.
The technology stack consisted of Firebase, Flutter, and GCP.
Ad Platform for Esports
Development of the advertising platform that connects streamers and advertisers. With the help of this solution, advertisers can turn on their
ads in the streams and streamers get paid.
Responsibilities:
• Project architecture.
• Team management.
• Testing and deployment of the project.
The technology stack included React, Java Spring, and GCP.
ads in the streams and streamers get paid.
Responsibilities:
• Project architecture.
• Team management.
• Testing and deployment of the project.
The technology stack included React, Java Spring, and GCP.
Product Matching Project
I developed a mixed fuzzy logic and NLP solution for matching product descriptions in the company's and its customers' databases. For this purpose, I implemented a data cleaning procedure and created a fuzzy logic matching procedure based on various metrics, such as Levenshtein distance and cosine modularity. In addition, I integrated pre-trained Bert models into the process. I was able to achieve high matching performance on the clients data.
Chatbox for Corporate Data Search
• Leveraged the power of GPT (like ChatGPT) to design a chatbot proficient in navigating through structured and unstructured company data, ensuring real-time responses to complex queries.
• Developed mechanisms to autonomously collect data from corporate sources, ensuring the chatbot is always up-to-date with the latest company information, solving the indexing problem.
• Set up advanced ML pipelines to process, validate, and integrate data from various sources, ensuring the chatbot's responses are consistent and reliable.
• Played a key role in co-developing the overall platform architecture, from designing an optimal database structure capable of handling a plethora of corporate data to creating an efficient API for potential integrations.
• Developed mechanisms to autonomously collect data from corporate sources, ensuring the chatbot is always up-to-date with the latest company information, solving the indexing problem.
• Set up advanced ML pipelines to process, validate, and integrate data from various sources, ensuring the chatbot's responses are consistent and reliable.
• Played a key role in co-developing the overall platform architecture, from designing an optimal database structure capable of handling a plethora of corporate data to creating an efficient API for potential integrations.
Wildlife Classification | Schedule Optimization
Developed a model for the detection and classification of wildlife for the environmental protection report and a model for optimizing driver shifts to increase the efficiency of material delivery.
Education
2013 - 2023
Progress Toward PhD in Computer Science
Goethe University - Frankfurt am Main, Germany
2010 - 2012
Master's Degree in Informatics and Applied Mathematics
Odesa Mechnikov National University - Odessa, Ukraine
2006 - 2010
Bachelor's Degree in Informatics and Applied Mathematics
Odesa Mechnikov National University - Odessa, Ukraine
Certifications
SEPTEMBER 2020 - PRESENT
Startup School
Y Combinator
JUNE 2015 - PRESENT
Machine Learning Summer School
Chalmers | Sweden
JUNE 2014 - PRESENT
Autonomous Learning
Summer School | Germany
MAY 2013 - PRESENT
Continuous Learning in Living and Artificial Systems
CITEC Summer School | Germany
Skills
Libraries/APIs
Pandas, NumPy, PyTorch, OpenGL, XGBoost, Azure API Management
Tools
PyCharm, ChatGPT, MATLAB, RabbitMQ, DeepSeek, You Only Look Once (YOLO), Amazon OpenSearch, AWS CloudFormation
Languages
Python, SQL, C++, Python 3
Platforms
Jupyter Notebook, Amazon Web Services (AWS), Docker, Blazegraph, Firebase, Azure, Databricks
Storage
Databases, Google Cloud, PostgreSQL, NoSQL, MongoDB, Redshift
Frameworks
Qt, Ionic, Laravel, Flutter, Flask, LlamaIndex
Paradigms
Scrum, Microservices, Dynamic Programming
Other
Machine Learning, Computer Science, Team Leadership, Artificial Intelligence (AI), AI Adoption, Strategy, Natural Language Processing (NLP), Regression Modeling, Neural Networks, Applied Mathematics, Web Scraping, Data Science, Optical Character Recognition (OCR), Image Recognition, Generative Pre-trained Transformers (GPT), Generative Pre-trained Transformer 3 (GPT-3), Large Language Models (LLMs), Chatbots, OpenAI, Chatbot Conversation Design, AI Agents, Machine Learning Operations (MLOps), Statistics, Neuroscience, Deep Learning, Recommendation Systems, Team Management, IT Systems Architecture, Technical Leadership, Concepting, Reinforcement Learning, Modeling, Computer Vision, Science, Startups, Fuzzy Logic, Autoencoders, Time Series Analysis, Signal Processing, OpenAI GPT-4 API, Pinecode, LangChain, Open-source LLMs, Large Language Model Operations (LLMOps), FastAPI, YOLOv5, Image Analysis, APIs, Ontologies, Model distillation, LoRa, Generative Artificial Intelligence (GenAI), Datasets, Meta Llama, Mistral AI, Groq, Tesseract, Dash, Amazon Bedrock AgentCore, Machine Learning Automation, Supervised Machine Learning, Capacity Planning, Integer Programming, Linear Optimization, LLM Agents
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