Technological tasks
The Tech-tasks module is tasks related to the development, implementation and improvement of new technologies, products and processes. If you have development, implementation and other needs you can post information below.
AI assistant for the admissions committee of a training center with CRM integration
We need an AI assistant (chatbot) for the admissions committee of the training center for additional IT education, which responds around the clock to incoming calls on WhatsApp, Instagram and on the website, helps to choose the direction of study and records for a trial lesson. Basic requirements: — answers to standard questions (directions, schedule, cost, format, addresses) in Russian and Kazakh based on a knowledge base that employees can edit without a programmer; — qualification of the application: student's age, direction of interest, convenient time, contact; — sign up for a trial lesson in the free slots of the schedule; — transfer of the lead and all correspondence to CRM, transfer of the dialogue to the live manager at the request of the client or in case of a difficult question; — analytics: the number of requests, the conversion rate for recording and attending a lesson, and frequently asked questions; — storage of personal data in accordance with the legislation of the Republic of Kazakhstan.
Decision acceptance deadline
13.10.26 (inclusive)
Preferred systems
Neurotechnology and artificial IntelligenceNumber of applications
5
Web service for automatic help generation at the educational center for additional education
We need a web service that, based on student data, automatically generates certificates and documents from the training center in PDF on official forms: certificate of education (for employers, government agencies, embassies, military enlistment offices), certificate of completion of the course, characteristics, petitions and letters. The employee selects a template, enters the data manually, or downloads a list from Excel/CSV and receives the finished documents one at a time or in batches. Main requirements: — Web interface with employee authorization and roles (manager, supervisor); — template library based on downloadable DOCX forms: new help can be added without editing the code; — manual input and mass import from Excel/CSV, document preview before generation; — upload to PDF one at a time and ZIP archive; — automatic calculation of dates and terms of training according to program settings; — a place for printing and signing, in the future — a QR code for verifying authenticity and signing EDS (NCALayer); — a log of issued certificates with numbers and dates; — import data from the accounting system (LMS) via API or upload; — storage and processing of personal data in accordance with the legislation of the Republic of Kazakhstan.
Decision acceptance deadline
13.10.26 (inclusive)
Preferred systems
Information processing and transformationNumber of applications
5
Ticketon Event Ranking System
To build a system that selects and arranges events for the user and context for each showcase display, adheres to business rules and proves the increase in purchases in A/B against the current order.
Decision acceptance deadline
13.10.26 (inclusive)
Preferred systems
Other technological solutionsNumber of applications
2
Development of a system for parallel filling of student applications on educational portals
To develop a server system for parallel filling of applications on educational portals based on verified data and documents of students. For each application, the system must run a separate isolated virtual environment with its own browser and an agreed network route. Main functions: transferring data to the form fields, uploading documents, checking the filling, saving drafts and tracking the status of applications. The operator should see the progress of the work, receive notifications about the need for intervention and be able to temporarily manage the selected session. CAPTCHA and login confirmation are performed with human participation. The final submission of the application is performed only after verification and confirmation by the operator. The pilot provides for working with three coordinated educational portals, support for at least 10 parallel sessions, and accuracy in filling in pre-matched fields of at least 98%.
Decision acceptance deadline
12.10.26 (inclusive)
Preferred systems
Разработка ПОNumber of applications
3
ML-forecasting and optimization of hourly deviations PLAN / FACT when working in the wholesale electricity market of the Republic of Kazakhstan
The aim of the project is to develop and implement a forecasting system for hourly production and/or consumption of electric energy to minimize deviations between planned and actual values when operating in the wholesale electricity market of the Republic of Kazakhstan. As part of the project, it is planned to create an analytical solution based on machine learning methods, statistical forecasting and optimization algorithms, which will generate the most accurate hourly forecast of energy consumption/output for the day ahead and, if necessary, update it as new data becomes available. The system must take into account historical hourly data, calendar factors, seasonality, day of the week, holidays and weekends, weather conditions, temperature dependencies, production schedules, technological modes of facilities, scheduled shutdowns, repairs, emergency events and other factors affecting the actual consumption or production of electricity. The main objective is to reduce the magnitude and frequency of hourly PLAN/FACT deviations, which lead to imbalances and financial settlement in the Balancing Electricity Market. The result of the project should be a decision support system that will allow: automatically generate a forecast for each hour of the estimated day; determine the expected deviation of the actual volume from the stated schedule; identify hours with an increased risk of significant imbalance; calculate the recommended target value, taking into account the projected consumption/output; compare several forecasting models and automatically select the most accurate one; implement regular retraining of models based on new evidence; display the indicators PLAN / FORECAST / FACT / DEVIATION in the analytical interface; evaluate the accuracy of forecasting and the potential economic effect of reducing imbalances. The key business effect of the project is to reduce financial costs associated with deviations in the BRE, increase the accuracy of daily planning and automate the process of generating hourly orders in the wholesale electricity market.
Decision acceptance deadline
07.10.26 (inclusive)
Preferred systems
MLNumber of applications
9
Development of software for centralized network management of stationary UAV detection and counteraction complexes "D2A PRO (16)"
Development of software for centralized management of the network of stationary UAV detection and counteraction complexes "D2A PRO (16)". It is necessary to provide a commercial offer.
Decision acceptance deadline
07.10.26 (inclusive)
Preferred systems
Intelligent control systemsNumber of applications
6
A system for generating realistic videos with a person and a specific product
Разработать и проверить прототип системы генерации реалистичных предметных видеороликов по фотографии товара и описанию сцены. Ищем разработчика или команду с опытом компьютерного зрения, генеративных моделей, видеосинтеза и автоматизации медиапроизводства. ОСНОВНОЙ СЦЕНАРИЙ Пользователь загружает фотографию конкретного товара, указывает его название и подтверждённые характеристики, описывает сцену и задаёт текст выступления либо тезисы. Например: «Человек на пляже берёт товар со столика, показывает его в камеру и рассказывает о назначении». Система подготавливает сценарий и последовательность действий, создаёт видеоряд и озвучку, синхронизирует речь и формирует готовый MP4. Перед генерацией пользователь может согласовать текст. Субтитры включаются отдельно. ТРЕБОВАНИЯ К ВИДЕО 1. Естественный визуальный стиль: правдоподобная кожа, мимика, взгляд, движения тела, освещение и тени; без заметных деформаций, мерцания и «пластикового» вида. 2. Точное воспроизведение товара: сохранение формы, пропорций, цвета, основных деталей, логотипа и читаемых надписей в согласованных ракурсах. Похожий предмет вместо исходного товара не принимается. 3. Реальное взаимодействие в кадре: человек берёт, удерживает и показывает предмет. Не допускаются прохождение пальцев сквозь товар, скольжение предмета в руке, самопроизвольное изменение формы и размера. Момент взятия и удержания должен быть показан непрерывно, а не скрыт монтажной склейкой. 4. Согласованная речь: естественная озвучка на русском языке и синхронизация губ. Характеристики товара берутся из материалов заказчика; система не должна придумывать свойства, сертификаты и отзывы. 5. Для пилота предлагаем ролики длительностью 15–30 секунд, вертикальный формат 9:16, экспорт MP4. Целевой размер — 1080×1920; исходное разрешение генерации и применение увеличения разрешения раскрываются отдельно. Окончательные параметры согласуются перед стартом. АРХИТЕКТУРА И НЕЗАВИСИМОСТЬ GPT, Claude или аналогичные модели предполагается использовать для анализа входных материалов, сценария и управления этапами. Исполнитель должен отдельно описать механизм создания видеоряда, движения, голоса и синхронизации губ. Допускаются самостоятельно развёрнутые модели с доступными весами, собственный рендеринг и обработка видео. Все компоненты должны иметь подходящие права использования. Аренда вычислительных ресурсов допустима; использование чужого сервиса генерации видео под видом собственной системы — нет. Обучение фундаментальной модели с нуля не является обязательным требованием. Базовый вход — фотография товара. Если для достоверного показа необходимы дополнительные ракурсы, размеры или 3D-модель, система должна сообщить об этом. По одной фотографии нельзя молча придумывать скрытые элементы товара и выдавать их за подтверждённые. ПРОВЕРКА ПРОТОТИПА На первом этапе необходимо доказать работоспособность ключевого сценария с пляжем и взятием товара в руку. Затем проверить переносимость решения на другие товары и сцены. Предлагаемый набор пилота — не менее трёх итоговых роликов с разными товарами и минимум двумя сценами. Набор примеров и критерии приёмки согласуются до начала работ. Исполнитель фиксирует все попытки генерации, время, расходы и ручные доработки, а не только показывает лучшие результаты. РЕЗУЛЬТАТ РАБОТЫ Рабочий прототип с простым интерфейсом или API; исходный код собственной разработки; конфигурации и инструкция по развёртыванию; перечень моделей, лицензий и внешних зависимостей; тестовые ролики и отчёт по качеству, времени и себестоимости одного принятого ролика. Заказчик должен иметь возможность повторить процесс без ручной работы исполнителя в сторонних видеосервисах.
Decision acceptance deadline
07.10.26 (inclusive)
Preferred systems
Neurotechnology and artificial IntelligenceNumber of applications
4
Linking completed physical volumes to BIM according to estimates and acts with tracing to finite elements
To investigate the recognition of estimates and acts of completed work and the comparison of their lines with the final BIM elements by area, type of work and volume. Create a prototype with two-way tracing: document and its version → line → type of work → Element ID and model version. For each relationship, save the quantity, unit of measurement, completion status, and matching basis. Select related elements in 3D by line of the document, and show the source lines and counted volumes by element. Maintain "many to many" connections, partial implementation, and the history of adjustments to documents and models.
Decision acceptance deadline
07.10.26 (inclusive)
Preferred systems
Neurotechnology and artificial IntelligenceNumber of applications
13
Digitizing the knowledge base into requirements and a graph of connections using AI
To investigate the extraction of verifiable requirements from standards, regulations, tables and diagrams. Create a prototype registry with the ID, source, version, owner, scope, and verification method. AI should offer links to standard solutions and dependent requirements, identify contradictions; publication of requirements and links is carried out after verification by an expert.
Decision acceptance deadline
07.10.26 (inclusive)
Preferred systems
Neurotechnology and artificial IntelligenceNumber of applications
12
Verification of design solutions according to corporate standards using AI
To investigate which requirements of corporate standards can be verified by geometry, parameters, and relationships of BIM objects. Create a prototype in which the AI interprets the requirement, and the approved rules perform verification. For a comment, show the standard clause, the affected elements and the proof; if there is a shortage of data, give the appropriate status.
Decision acceptance deadline
07.10.26 (inclusive)
Preferred systems
Neurotechnology and artificial IntelligenceNumber of applications
7
