Natural language processing and
machine learning

laboratory

мы говорим по-русски // english
For questions: iwitaly@potehalabs.com
We conduct production ready research and development to solve natural language understanding and classical machine learning problems.
Our focus is dialogue systems, named entities recognition, general text analysis and processing, recommendation systems and statistical predictive models.
Our range of work from the task description to the deployment. We help our clients pick out opportunity from the problem and decide on how to resolve it.

Work & Stack

WE USE Python 3.6 + Postgres + Django/Flask + all that's needed. For deployment we prefer Docker containers with orchestration tools like ECS. In machine learning we prefer classical tools such as Sklearn + Keras + Pytorch
our work process
// Clarifying all technical features of the problem
// Inventing metrics, finding out the details and studying the side cases. Providing customer with a solution plan
// Data cleaning and preprocessing
// Study of experiments (theory) with data
// Implementation of baselines and experiments with models
// Checking the quality on the dev environment of models
// Writing an API for models, Preparing for production and Deploing microservice including docker
we available for call or messages 24/7

// keep our work as transparent as possible for our clients
// give full access to our repos and task boards
// Once a week we prepare a talk or presentation about sprint results and give it in person

Projects

Dodo Pizza // №1 Pizza in Russia
What we’ve done: recommender system, classical machine learning
Mercuryo // Cryptowallet for Everyday Use
What we’ve done: Neural networks, face and documents recognition
All projects 4

Work & Stack

WE USE Python 3.6 + Postgres + Django/Flask + all that's needed. For deployment we prefer Docker containers with orchestration tools like ECS. In machine learning we prefer classical tools such as Sklearn + Keras + Pytorch
our work process
// Clarifying all technical features of the problem
// Inventing metrics, finding out the details and studying the side cases. Providing customer with a solution plan
// Data cleaning and preprocessing
// Study of experiments (theory) with data
// Implementation of baselines and experiments with models
// Checking the quality on the dev environment of models
// Writing an API for models, Preparing for production and Deploing microservice including docker
we available for call or messages 24/7

// keep our work as transparent as possible for our clients
// give full access to our repos and task boards
// Once a week we prepare a talk or presentation about sprint results and give it in person

Reviews

Dmitry Garin
// Digital Transformation Director at Rybakov Foundation

We asked Poteha for financial data analysis as a stage of the due diligence we conducted: they had to quickly go into the data from several sources, check their correctness or find suspicious transactions and make a forecast. Even under tight deadlines the analysts were very careful and attentive, the thoughtful work of the team is felt. Guys participated in our discussions a lot, provided explanations to everything found in data and quickly answered all our questions. Will be happy to work together again!

Sergey Golovanevsky
// CTO & co-founder at AmoCRM

AmoCRM is developing as a messaging-powered CRM and we wanted to add options on chatbots and autoanswering in form of a lightweight instrument for our clients to communicate with their customers. Little data for training was the task feature which motivate guys to come up with nontrivial architectures and methods of text augmentation. We conducted tens of experiments with changing conditions and generated datasets. Poteha’s solution was announced in our San Francisco release and then added to the Russian amoCRM version.

Sofia Kapkova
// Founder & CEO at CDK and Nonfiction.film

Nonfiction.film is a young and ambitious project. Therefore, for its programming and development, we attracted the same young and daring. We learned a lot while working on the project together. The team suggested a lot of complex modern approaches to designing a new platform, which allowed to create a resource for the most demanding user. Further development of the platform we will definitely delegate them.

Dmitry Barbanel
// A graphic designer and a partner in designworkout, masterskaya.pro

Poteha Labs is a company of young superprofessional guys, who provided Masterskaya with a breakthrough in serious web projects. Thanks to their support and enthusiasm, in August 2017 the educational project DesignWorkout started in Telegram. Nobody did this before and, apparently, has not done so far! The project brought together hundreds of creative designers and defined a new direction for the development of educational projects of Masterskaya, including Campus.

Daria Rodionova
// Head of User Research at Toadman Interactive

Poteha helped Toadman Interactive with analytical system backend. Guys at Poteha showed a great deal of care and professionalism when It came to Toadman’s business goals. They are definitely experts in their field and I am happy to recommend their services. Looking forward to working with them again!

Dmitry Zaruta
// CEO at EasyTen

I really enjoyed working with guys from POTEHA. First of all, I would like to note the team responsibility and the desire to understand the product, and not just perform the assigned tasks as is. They have great experience in developing both from the technical and business side. Together they have built a flexible and scalable analytical system, as well as many other tools for the back office and server side of the mobile application EasyTen. I recommend them as a team that is able to solve each problem they ever face.

Greg Waisman
// COO, co-founder at Mercuryo

Guys provided us with a custom KYC system for faces and documents recognition and comparison, based on computer vision. The result of their development is rapid, performant and production ready. Speaking about the team’s work process, I appreciate their accurate in time work in 1-week sprints, collaboration transparency and overall diligence. Thanks a lot!

Alexander Andronov
// CTO at Dodo Pizza

In Dodo Pizza we’ve been gathering data related to sales, pizza making and delivery, customer activity since 2011 and now we want to make a ‘data breakthrough’ so that we could base our decisions on it. To accomplish this we need highly experienced developers and Poteha Labs team precisely meets our demands. Together we have to build a lot from scratch, develop architectures and lay the foundation for the analysis. It requires deep understanding of our product and marketing strategies. And I rarely met someone else who would dive into issues as fast as Poteha team. We prefer not to delegate our partners narrow tasks: we only describe problems in general and expect the solutions that they offer themselves to help us move a lot forward.

Call +79261061040 or write us
We work with research-friendly companies, which understand risks and benefits from machine learning R&D.
Describe your project idea and write a company name
if any
Email or phone
We reply within 48h
Contact sales
We work with research-friendly companies, which understand risks and benefits from machine learning R&D.

We are looking for cool engineers

Fullstack developer
Engineer in ML
Technical Project Manager
We are looking for accurate developer. If you do your work until the end and understand how the design works, if you want to grow and evolve, please, contact us! We offer fascinating projects, pay average market rate, and give you opportunity to work from anywhere you would like. Our team consists of engineers and data scientists, the average employee age is 24. You are expected to know adaptive layout, React, Redux, Node or Python, database. At the interview we would like to learn about your previous experience and discuss possible architectures of web projects.

You are expected to have:
// HTML, CSS and JavaScript knowledge

// experience in Django

// the ability to synchronize different parts of code

// basic knowledge of computer science

// experience in PostgreSQL, MongoDB and other DB

// knowledge of Gitflow

// knowledge of adaptive layout

Fullstack developer
Engineer in ML
Technical Project Manager
SEND YOUR CV
We’re also creating ML-based web and mobile services // potehadevs.com
Contact sales
CLOSE
Describe your project idea and write a company name
if any
Email or phone
We reply within 48h
Contact sales
We work with research-friendly companies, which understand risks and benefits from machine learning R&D.