1. Product Development Service

Machine Learning Engineer

Artificial Intelligence is a key towards revenue growth and understanding customers. Many day to day tasks gain substantially with machine learning applications.

Ability to classify an eye disease at an earlier stage, perform a risk assessment for loan applications, find a similar item in your e-commerce platform, identify an intruder into a PC based on keystroke and mouse dynamics, help with image editing task by segmenting people, dogs, food, or evaluate student essays automatically across wide range of topics, or automatically transcribe a podcast with your android phone in offline mode, or parse text information in PDFs and make them searchable, or do style transfer on e-state imagery to increase attractiveness, these and many other tasks were completed successfully to bring more value into already built products.

Please, contact, if you need something better!



Artificial Intelligence, Deep Learning, Machine Learning, Computer Vision,


Image Processing, Face Detection, Face Recognition, Natural Language Processing, Automatic License Plate Recognition, Calibration, Object Detection, Automatic Speech Recognition, Image Cropping, Semantic Segmentation, Tracking, Statistical Analysis, Data Engineering, Data Analytics, Scientific reports, Human Pose Estimation, Numerical Analysis, Data Mining, Classification, Clustering, Nearest Neighbor Search, Time Series Analysis, Optimization, Regression, Forecasting, Statistics,

Tesseract, Google Cloud Vision, OpenCV, Dlib, OpenALPR, Rasa, Chatfuel, DialogFlow, DeepSpeech, Wit, Google Vision API,


Pytorch, Python, TensorFlow, Scipy, Keras, PySpark, Pandas, Numpy, Scikit-Learn, SKLearn, Jupyter Notebook, IPython, Matplotlib, Matlab,

Hadoop, Kafka, Spark, HBase, HDFS, Redis,


Alexa, Google Assistant



Big data

Neural Style Transfer, Transformers, Deep SORT, Simple Object Real-Time Tracking, Alphapose,

2. Company Structure Visualization


Product Idea

I am looking for someone who could build an App which would essentially extract data from Xero HQ for the purposes of creating a client structure ie. Company, Trust and details thereof including Directors shareholders, beneficiaries etc.3

This link will give you an idea of what I am looking for: https://www.account-kit.com/features — see client map under the Consulting tile.


  1. account-kit
    some financial tool, likely to help with paying taxes, and keeping track of payments
  2. Xero HQ
    yet another clients management software, as well as payments management, has rest api, samples of code on github

Development Roadmap

  1. 07.06.2021
    1. updates
    1. reviewed account-kit and xero hq
    2. added rest api for xero-hq
    3. need to research company structure visualizaton ready-to-use tools
    4. need to retrieve some sample data from Xero API, since it’s not clear what is present inside.
    2. 9:30–11:11am (gmt+3) 1h11m
  2. 07.06.2021
    1. updates
    1.1. related papers
    1.1.1. https://arxiv.org/pdf/1901.07411.pdf
    1.1.2. https://arxiv.org/pdf/1209.1479.pdf
    1.1.3. https://arxiv.org/pdf/1703.00485.pdf
    2. 10:30–11:30pm (gmt+3) 30m

3. Project Management Software


Product Idea


Build A Project Management Tool Like Asana and Trello

List of features: https://asana.com/pricing


  1. Asana
    project management software, has rest api, can be used if anyone is willing to redesign asana UI
  2. Trello
    yet another project management software, as well, has API
  3. OpenProject
    5k on github, open source alternative to trello and asana build with Ruby

4. Text Summarization


Product Idea

We are looking for someone that can create a NLP process on one of our development systems using Remote Desktop Connection. If you search for “Text Summarization Approaches for NLP — Practical Guide with Generative Examples” at machinelearningplus{dot}com you can see what I mean.

Requirements on the framework you will use is that it is open source and runs locally on a development system.

The process should be able to run on either CPU or GPU or both.


Is to extract text summerauzations in correctly composed english sentences regarding available keywords (e.g. architect, build date, height, number of visitors per year) found in the source texts from several sources e.g. wikipedia source and others.

steps invloved: (should all be automated)
* based on a subject e.g. eifel tower. Take a frensh wikipedia page and other french information regarding that subject.

* Remove any non translatable and non relevant elements.

translate the sources to english.

* split the sources into paragraphs.

* combine paragraphs from both sources grouped on sub topic

* generate summerisations based on keywords. Result should be coherent english without duplicate or similar sentences

* translate the result to spanish and chinese using a good quality translator like google translate or better.

* store the results in 3 files on disk with a specific file name.

Development Roadmap

1. text summarization models

1. https://github.com/miso-belica/sumy
sumy lex-rank --length=10 --url=http://en.wikipedia.org/wiki/Automatic_summarization
a library for text summarization, lex-rank prints out most relevant sentences from the text.
lex-rank, lsa, lunh, kl-sum (all of the methods should work fast on CPU, but not that good as NN based ones)
wikipedia fetching is based on requests and some HTML parser in sumy
2. NN abstractive summarization
both models has an example of summarization inference
gpt-2 doesn't have an example of summarization directly
same for XLM as GPT-2
in general the question is what are RAM requirements, and CPU inference speed in case models are being used on some cheap cloud.

2. what's missing

1. some translation components, french<->english, english<->spanish, english<->chinese
2. aspect based summarization, like skewed towards a specified keyword
3. some ready-to-use UI for the app
4. proper estimates on the system throughput and deployment costs

5. HTTPS stack for Atmel (ATSAMD51J20)


Product Idea

Need a working source code for HTTPS server and client service running on ATSAMD51J20 series microcontroller.

The HTTPS server and client source code itself must not rely on Atmel ASF or Atmel Start project.

The rest of the project source, including microcontroller IO, initialization, and sockets library may be based on either barebone or Atmel ASF/Start source.

The project cannot be written using FreeRTOS or any RTOS.

HTPS source must be based on State machine and callback logic.

Development roadmap

1. links

1. https://github.com/BriscoeTech/Arduino-FreeRTOS-SAMD21

2. https://learn.adafruit.com/adafruit-all-about-arduino-libraries-install-use/library-manager

3. https://learn.adafruit.com/search?q=compile

4. https://learn.adafruit.com/category/adafruit-io?guide_page=2&total_count=71&total_verbiage=total+series

5. https://learn.adafruit.com/building-and-running-micropython-on-the-esp8266/build-firmware

6. https://github.com/adafruit/circuitpython/issues/2202

7. https://github.com/khoih-prog/EthernetWebServer_SSL#installation
looks pretty good, from scratch http and tls stack. need to figure out what way to compile firmware with it and flash onto AdaFruit board

8. https://forums.adafruit.com/

9. https://github.com/adafruit

10. https://www.freertos.org/mqtt/mutual-authentication-mqtt-example.html

11. https://docs.aws.amazon.com/embedded-csdk/202103.00/lib-ref/libraries/standard/coreHTTP/docs/doxygen/output/html/index.html#http_memory_requirements

12. https://github.com/aws/aws-iot-device-sdk-embedded-C

13. https://github.com/aws/aws-iot-device-sdk-embedded-C/tree/main/demos/http/http_demo_basic_tls
has http and tls, but reliaes on freertos

14. https://github.com/FreeRTOS/coreHTTP/tree/b349f89e67f5feef51438dc09ed89959dedc18d5

15. https://github.com/FreeRTOS/FreeRTOS/blob/main/FreeRTOS-Plus/Source/Application-Protocols/network_transport/freertos_plus_tcp/using_wolfSSL/using_wolfSSL.c

16. https://www.freertos.org/network-interface.html#examples

17. https://www.freertos.org/network-interface.html#examples

18. https://www.freertos.org/network-interface.html

19. https://www.freertos.org/application-protocols.html

20. https://github.com/FreeRTOS/coreHTTP

21. https://github.com/FreeRTOS

22. https://nitk-nest.github.io/

23. https://github.com/x893/SAMR34

24. https://github.com/xythobuz/avrNetStack

25. https://fnet.sourceforge.io/

26. https://github.com/embedded-office/canopen-stack

2. well in general there’re some ready-to-use librarie. need to compile it and test on the device. not clear though what kind of product can be built. by default adafruit board is pretty general. It’s likely the device is to be used as a part of some IoT hardware.

To get a complete product, would require some hardware device design, and a clear justificiation what for to apply not FreeRTOS based solution.

6. Real Estate Accounting


Product Idea

Real Estate Office. dealing with multiple agents.

accounts payable function

accounts receivable function

bank reconsiliation for 3 accounts

Development Roadmap

1. links

1.1. https://github.com/etewiah/property_web_builder
pwb is an example of community+enterprise version of a real estate office
web site builder. Has good examples on pricing, UI/UX layout, and features.

1.2. https://github.com/easypropertylistings/Easy-Property-Listings
yet another example of ready-to-use plugin, this time in PHP,
integrated as a plugin for wordpress, has pricing, features, UI/UX samples.

1.3. https://github.com/etewiah/awesome-real-estate
a typical awesome list, useful to revise for more relevant info on the field.
2. well, there’re some competitors, bloggers, ready-to-use github projects
didn’t review in detail payment functions, as well as banks integration.
but to start with should be enough to be build a prototype.
maybe makes sense to consider template development from scratch costs for the website.
for a software developer, there’s not much to do here without a detailed conversation with a customer

7. Printing Service with Color Density Control Feature


Product Idea

Create tool to change color density of prints on the print server..

I need a tool that reduces the density of prints, where each printing made on the print server is subjected to a decrease in its density, in proportion to previously set values. It needs to be customizable.

Development Roadmap

1. links

1.1. https://github.com/google/cloud-print-connector
1.2. https://github.com/OpenPrinting/cups
1.3. https://qz.io/
it’s likelty that a customer has desired to customize qz tray application directly.
1. https://github.com/qzind/tray/search?q=color+density
2. https://github.com/qzind/tray/blob/561bdeaccd9036b010672f6503ec9a01b917aae1/src/qz/printer/PrintOptions.java
3. https://github.com/qzind/tray/blob/802d1f8f2d47b7c36d9f6c1187a117d0170f6567/src/qz/printer/ImageWrapper.java
4. https://github.com/qzind/tray/blob/802d1f8f2d47b7c36d9f6c1187a117d0170f6567/src/qz/printer/action/PrintHTML.java
5. https://github.com/qzind/tray/blob/802d1f8f2d47b7c36d9f6c1187a117d0170f6567/src/qz/printer/action/PrintImage.java
6. https://github.com/qzind/tray/blob/15a7e38c4c0e5df4673c71a5368dc7f38bfa5190/js/qz-tray.js
in gneral qzind/tray has density configuration option.
So its either about underlying printer driver issue,
or some general misconfiguration of the tray software.

8. Custom game engine to make games similar Magic Arena


Product Idea

Brief Project description

Implement a generic and extensible turn based card combat game engine with look and feel

similar to Magic Arena .

Ready to implement fantasy turn based games.

The goal is making an engine capable of doing games with similar board, cards interaction and

effects with some nuances. For example, do not need 3D models, only board view.

To test the engine, develop a prototype with the following features:

* Splash screen

* Main screen with background and game menu

* Classic options menu

* Showcase of a battle ( no IA, just test rendering, effects, send card to cemetery, etc).

Technical Specifications:

• Language C/C++17.

• Renderer OpenGL 4.4

• Libraries SDL2, GLM.

• Build system Cmake.

• Target plattforms Windows / Linux (Debian)

Deliverables through gitlab repo

• Game engine Source code

• Class diagrams

• Prototype Source code

• Builded Prototypes Windows and Linux

Source code specifications

• Well designed and commented in english when needed

• Pattern oriented when allowed

• Easily modificable and extensible

Game Engine features

• Multithreaded

• Decoupled subsystems, specially rendering and game logic

• Basic Physics

• Basic Networking

• Developer console

• Chat console

Game Engine Subsystems

• Renderer

• Input

• Audio

• Networking

• Physics

• Resource Management.

Development Roadmap

1. links
1. https://ycarowr.itch.io/cardgameui
a demo of cards UI, pretty basic, no game logic, based on unity
2. https://github.com/ycarowr/uicard
engine for the demo above, written in C#, is likely to be based on Unity.
3. https://github.com/manuel-777/mtg-arena-tool
Magic Arena analysis app, has a proper UI/UX. Need to research underlying algorithms.
As well as see what way it binds to Magic Arena game itself,
and what are the import/export options.

4. https://github.com/magefree/mage
a good open source implementation of Magic Arena like game, has rather ugly UI.
But underlying implementation has client/server, AI based bot.

5. https://itch.io/
a catalog of games, good to research UI/UX, features examples
6. https://magic.wizards.com/en
a good place for understanding the game itself
2. In general there’re some related parts to make a copy of the game. Require some more detailed development planning, since technologies are different for various parts. Maybe, its likely the development is to happen from scratch. Depends on the project budget. But to build a proof of concept within 1–2 weeks should be enough.

9. IoT sensors powered by Zephyr RTOS


Product Idea

Zephyr RTOS and Device Tree Expert

This job requires someone who has deep knowledge of the Device Tree,

especially the one which is used by the Zephyr RTOS.

Development Roadmap

1. links
1. https://docs.zephyrproject.org/latest/guides/dts/howtos.html#troubleshoot-devicetree-issues
low level documentation on device tree configuration within Zephyr RTOS framework in C.
2. https://docs.zephyrproject.org/latest/search.html?q=device+tree&check_keywords=yes&area=default
a broader set of documentation on the subject, may research later on.
3. https://github.com/zephyrproject-rtos/zephyr
Apache 2.0 Licensing, should be ok for proprietary products.
supports quite many boards, is likely to be used for https project as well.
4. https://www.zephyrproject.org/the-rtos-for-the-future-of-the-iot-laird-connectivity-and-the-zephyr-project/
some more links on zephyr rtos, it’s likely it has been introduced in 2018
2. makes sense to discuss a particular project. since Zephyr is a general purpose framework for embedded devices. Among issues, is that most of the already present devices on the market should have a huge code base which is differernt per each device, and some industries may depend on it. That’s why porting everything to Zephyr can be idealistic. But as a single place of architecture ideas, and communication platform for open source IoT develoeprs. That’s good.

10. Voice recoginition system with AI/ML


Product Idea

Specification as below:

1-user record his voice (read a spicific message for first time)

2-voice recording will be saved in database as reference with time/date

3-AI/ML to check the voice recording

4-That recording will be used for user authnetion and verification (new message will be displayed) and user need to say it, and code must know this is same user or not.

5-Backend aowkr with API only

Development Roadmap

1. links
1. https://fosspost.org/open-source-speech-recognition/
1. https://github.com/mozilla/DeepSpeech
2. https://kaldi-asr.org/
3. https://github.com/julius-speech/julius
4. https://github.com/flashlight/wav2letter
5. https://github.com/PaddlePaddle/DeepSpeech
6. https://github.com/NVIDIA/OpenSeq2Seq
7. https://github.com/pytorch/fairseq
8. https://alphacephei.com/vosk/
9. https://github.com/athena-team/athena
10. https://espnet.github.io/espnet/
there’re quite many speech-to-text open source components to use.
an alternative is to integrate REST API based services.
Quite many companies provides this services for a quite a while.
2. Need to add:
1. voice identification component
2. some general solution architecture which includes all of the backend and frontend components
3. need to add some UI/UX examples to rely on
3. In general the task by itself is not good for a startup like product. That’s why it makes sense to discuss a prticular use case. Since quite likely just integration some rest api should solve the issues.

Freelance for artificial intelligence domain