Reading Time: 15 minutes

Introduction

There are a lot of AI Companies that offer Machine Learning development. Even choosing the right ones for your project could take a decent amount of time, not even mentioning to pick the one that fits.

 

During 2017 the search for Artificial Intelligence Companies increased four times compared to the previous demand:

 

Here at SPD Group, we offer Machine Learning and Artificial Intelligence solutions including Predictive Maintenance, Predictive Analytics, Virtual Personal Assistant, Cloud-based AI models, and Raw data management. We developed a predictive maintenance solution for 40 000+ payment terminals and NLP-based data clusterization solution for financial data and software company. If you are interested in Machine Learning development, have any questions and suggestions, or you want to start your ML project, feel free to contact us.

 

We decided to help CEOs, established companies, and startups that are looking for AI & ML Developer to save time by going in-depth with top AI Companies in the industry. The focus is on Machine Learning development, and we go beyond basic Clutch.co descriptions to provide you as much insight we can.

 

To do this we got help from Head of RnD Department Dmytro Kuriatnyk, ML Tech Lead Oleksandr Melnyk and Project Manager Olena Kovalenko. They are a part of the SPD Group and have practical experience in this field. We collected as much information as we could and assessed companies by the following criteria: Response Time, Expertise, Long-Distance Engagement, Management, Types of Engagement and Pricing, Maintenance and Scaling. Each point is evaluated on the scale from 1 to 5, and that’s how we form the overall average rating that is displayed with the name of the company.

 

So here is our list of AI Development Companies. Hopefully, it will help you to find your perfect match!

Machine Learning development meeting

 

AI Company from London, UK – 4,3

This is an award-winning consultancy from the United Kingdom that specializes in Machine Learning, NLP and Data Science. They provide full lifecycle consultancy to bring AI to your business. You will talk to the CEO/co-founder that responds in less than an hour.

 

Expertise
They rely on Python, Google Cloud, AWS as well as other infrastructure services. Offering Senior and Junior Data Scientists with Ph.D. degrees. Using mainstream AutoML models for fast prototyping, they do the bespoke modeling and feature engineering for higher results (accuracy, precision, recall, (F1)).

 

Their website contains an exclusive AI/ML expertise description. The services are:

 

  • Data Extraction
  • AI Development
  • Consultancy
  • AI Workshop

 

Among the case studies, there are 10 that implemented AI/ML in the following industries:

 

  • E-Commerce
  • Financial Services
  • FMCG
  • Insurance
  • Manufacturing
  • Marketing
  • Recruitment
  • Retail
  • Technology

 

They have an engine that extracts data from web applications with the open RESTful API.

 

Communications
They claim to have good communication from remote locations due to a distributed team with offices in London and Portugal. For the conference calls and daily standups, they use Zoom.us. Travel expenses, including hotels, and food, are paid by clients, you can set a fixed travel budget for 1 meeting per month.

 

Management
Starting a project with this partner you set acceptance criteria that both parties agree on. Sometimes they need to suspend the project for 1 or 2 weeks to collect data for retraining the models. They use Trello for project management and Github for code control and updates. The company assigns a project lead to each case. They run a TDSP management process for Data Science project management. It’s quite similar to an Agile method but enables re-tune the models and iterative approach. You can learn more about it on their website. The development includes 6 stages. The PoC phase usually lasts for 3 months, the scoping could be done in 1-2 weeks. The duration of the other stages depends on the size of a project. You will be able to access the CEO, Project Lead, Data Scientists, Dev Ops, and Software Engineers during the development process. Having the two offices and some remote staff they say it won’t be an issue for the final result.

 

Types of Engagement and Pricing
The pricing model is Time & Materials, plus a technology license for all third and first-party technologies they use. They presented their daily rates. To make this process easier there is a payment in lieu after 30 days. Meaning they work for 30 days and then invoice for that.

 

Maintenance and Scaling
After the release, they provide managed services that include retraining models based on evaluation criteria that they can monitor. In the first year, they do this 2 days per month for free. Training of the system or training to improve models if necessary are also provided.

 

Advantages

  1. The response of the CEO/co-founder was quick
  2. The opportunity to meet the team, daily standups and set budget is good. But they were unclear about the way they deal with different time zones.
  3. The usage of specific methodology instead of simple Agile is a plus and not the case for all companies in our list.
  4. Free of charge models retraining 2 days a month, for the whole year.

 

Disadvantages

  1. Access to all company representatives involved in the development process.
  2. No Fixed-Cost payment model, which will be uncomfortable for some clients
  3. Trello is a quite basic tool, and not even for IT projects in particular, they can use something more professional.

 

AI Company from Kyiv, Ukraine – 3,3

 

Since 2005 they provide professional software development services for the sustainable growth of SME and startups. This company is located in Ukraine and have 50% focus on Artificial Intelligence development. It took 30 minutes to receive a reply from the CEO!

 

Expertise
While the cases are under strict NDA agreement, they were able to mention Predictive Maintenance for Coffee Machines Case and Prediction Tool for Supply Chain that achieved 85% accuracy. More use cases were attached in the presentation.

 

Communications
Remote management is their sweet spot, as they say, being aware of the cultural aspects of business collaboration and modern methodologies. Working with companies all over the world from different periods of time they could mention Anagog (Israel) for 5 years, Soraa (USA) for 2.5 years, SoSitter (Canada) for 3+ years. You will be always in touch with the Account Manager and have weekly sync ups with a Project Manager. The average response time is less than 4 hours, using an omnichannel communication environment: Email, Slack, Telegram, WhatsApp, Viber, Hangouts, Skype, Zoom, direct phone calls, and any other tools. Meetings in-person are also being practiced, for example, they are doing 3 days workshop in Singapore now. A client pays for all travel expenses including working hours, per diem, travel costs, accommodation costs and visa costs. The client also free to visit their office in Kyiv.

 

Management
To form a team Project Manager/SCRUM Master in close collaboration with the Staffing Department carefully assesses each team member. They stick to the SCRUM management model in software development. To ensure that the core team will stay until the very end fair salary and bonuses, possibilities to grow, training and team building, meetings with a supervisor are being practiced. Normally, communication with the Project Manager and Account Manager is suggested. However, you will be granted access to any team member you want to, including developers, QA & BA, PM, AM, CEO, etc. The core team (50%+) is located on the Kyiv office, part of the development team is distributed but it won’t affect the deadlines and budget due to the work of Scrum Masters. They admit that sometimes they actually fail to meet the deadlines, but this happens due to the Change Request, clients delay or operational risks. If the problem is on the developer’s side they raise a red flag to the client ASAP, investigate the issue, develop a viable plan and pay for all extra expenses caused by this issue. The average recruitment time is up to 5 business days.

 

Types of Engagement and Pricing

 

There are two main models of cooperation:

 

  • Managed Project services, when the company take care of the project from A to Z
  • Dedicated Team services, when the company augments your in-house team with remote talents

 

The T&M model is recommended in order to provide you with the maximum level of flexibility. Billing and invoicing are fully transparent and client-oriented. The company sticks to USD but is flexible to adjust to other currencies and accept payments via alternative means of payment.

 

Maintenance and Scaling
According to the attached case studies, all developed projects are supported after the release. Many of the clients engage in another project with this company. However, there is no information on how exactly they support a project and what are the scaling options.

 

 

Advantages

  1. They are offering a big list of services, and the diversity of ML expertise in different industries is impressive
  2. They can also work with Big Data
  3. Cooperation with clients in different time zones
  4. Covering expenses if the problem will appear from their side
  5. Direct access to every team member

 

Disadvantages

  1. Quite a few technical details on their website
  2. No daily standups
  3. Response time is less than 4 hours, it is quite long
  4. The rates for their work aren’t mentioned
  5. Hard to evaluate how they scale projects and support them after release due to lack of information

 

AI Company from Minsk, Belarus – 3,9

This company from Belarus provides a full-cycle development of comprehensive AI-powered solutions since 2012, delivering for companies such as Toyota & Lexus, TL Nika, Dragon Sourcing, MyMediAds, Boxforward. Head of Business Development will reply to you in less than an hour.

 

Expertise
They claim to implement ML in several projects. Including IoT project, a module of suggestions and cross-sale for an eCommerce website, ML-based app for sports predictions. They can’t give more details without signing NDA.

 

Communications
Usually, they hold daily “standup” calls and every 1-2 weeks showing the demos. But this can be adapted to the needs of a client. They use Slack, Skype, Hangouts, along with Jira and Confluence for project management. There are two types of reports on the projects: Monthly/Bi-weekly Reports and Real-Time Reports in Jira Atlassian. Travel expenses are covered by clients.

 

Management
They are located in one office. The team is formed based on the stage the company starts the engagement, required expertise, timeline, and available resources. They don’t assign developers for a few ongoing projects but mitigate the risks of developers turnover. The Project Manager is the key contact for all questions and the Unit Coordinator will handle everything else. You can meet the whole team in Minsk for tech-related questions and also CEO and CTO.

 

Types of Engagement and Pricing

There are two models available Time & Materials and Fixed Cost. T&M is based on the actual cost of direct labor specified at hourly rates, an actual cost of materials and equipment based on their 2-week or 4-week reports. The Fixed Cost model is based on well-defined and written requirements performed by a Business Analyst. All payments within this model are made on pre-defined milestones. However, all payment terms are discussable. You can only pay via bank transfer, no other options are available.

 

Maintenance and Scaling
There are a few ways this company can help after release:

  • They offer a free warranty in the case when they planned, developed, tested and released a functionality. If they are responsible for the development only, for example, they will fix bugs on basic hourly rates.
  • This partner will stay with the client even after the warranty period is over, you can discuss the terms of it.
  • Knowledge transfer is always provided in every project. You will receive all the necessary documentation.

 

 

Advantages

  1. Daily standups and weekly demos are nice features. They are flexible in communication and management tools
  2. 2 payment models: T&M and Fixed Price
  3. The warranty period and the knowledge transfer

 

Disadvantages

  1. Machine Learning is not the primary focus of their business, the majority of projects were made on Ruby on Rails, they have mobile apps on Android/IOS, projects with the usage of AWS and Heroku. ML technologies aren’t even in the lists of their technical stack for the projects.
  2. Regardless of sufficient details of the development process, and the fact that the company gives dedicated developers. There is no access to the developers directly, only C-level.
  3. Payment is only available through the bank transfer.
  4. Scaling is unclear

 

Artificial Intelligence development

 

AI Company from New Jersey, USA – 4,5

Providing design and development services from headquarters in Livingston, New Jersey and R&D center in Grodno, Belarus since 2001. They help insurance, life sciences, recruitment, media and companies in other industries. You will talk to Vice President with response time less than two hours.

 

Expertise
The PDFs of three case studies were presented: Semantic Search Engine, Cloud System for Document Digitization and Road Pothole Detection with Computer Vision. The most common technologies used are TensorFlow, Scikit Learn, Tesseract and Keras.

 

The full tech stack:

 

  • Programming Languages: Java, JavaScript, Ruby, Python, PHP, Scala, PL/SQL, Objective-C, Switch
  • Databases: Oracle, PostgreSQL, MySQL, MongoDB, SQSQL, SQLite, CassandraDB, Sybase, Firebase
  • Frameworks: Spring, Hibernate, Ruby on Rails, React, Native, Angular, Vue, Node.j.s., Symphony, Yii, Laravel, jQuery, Apache Spark, TensorFlow, Scikit Learn
  • Platforms and Solutions: Web, iOS, Android, J2EE, IoT, Liferay, Tibco BPM, Drupal, Paypal, Stripe, AlphaCard, ProPay, Square

 

Communications
This company has customers all over the world, some of them are from the USA, so they have experience working remotely with 7-8 hours of time lag. Usually, they run daily/weekly meetings with customers using Skype, Hangouts or any other software. The whole team is available by Email, Skype or Slack. Business trips are paid by the client, but all expenses are being discussed beforehand.

 

Management
When the project starts Project Management will form a team, but they admit there are times when not all team members make it to the end of development. You will have access to the team, as well as the CEO. They admit risks and situations when they fail to meet the deadlines and make the budget. The reasons are tasks that appear to be harder than estimated and new functionality added by the client. To prevent this they suggest frequent communications, but when something like this occurs they promise to find the solution that makes the client happy. To start a project they need to analyze your data samples to make an offer. After working out all the details an agreement is signed. Then the team is formed in about 2 to 4 weeks, a kick-off meeting is being held and development starts.

 

Types of Engagement and Pricing
They provide Time & Materials and Fixed Cost models with a detailed guide on them on their website. They provided monthly rates for hiring experts. There is a 15% discount if you hire 3 specialists for 3 or more months. They eliminate bureaucracy and paperwork and are open to discuss other options.

 

Maintenance and Scaling
You can increase the scope of the project during the development process, all changes will be regarded as Change requests and will be estimated additionally. There is a 2-month guarantee period with a free bug fix. After this period the Service Level Agreement is signed covering all future maintenance processes. There are several clients with 24/7 support.

 

 

Advantages

  1. They are honest to admit the typical situations with some fails of deadlines and client requirements
  2. Transparent pricing and even discounts
  3. 24/7 support and 2 months free period of bug fixes

 

Disadvantages

  1. They provide Software Development in general. Considering the lack of AI/ML information on their website, they are just getting started in this field.

 

 

AI Company from Hanoi, Vietnam – 4

This company from Vietnam delivers software development services with quality acknowledged by Fortune Global 500. You will speak to the co-founder that responds in two hours.

 

Expertise
They will send you a PDF file with a variety of different technologies they use for AI/Machine Learning/Big Data projects. They claim to have 25+ projects in the AI field.

 

Communications
They have a lot of clients in Europe and are comfortable with 4-5 hours overlap. For medium to big size projects, they follow Agile methodology with 2-4 weeks Sprints with frequent communication with the client on each stage. They prefer to use Emails and Microsoft Teams for communication. As for travel expenses, their experts will fly to Europe while you will pay for tickets and daily expenses. There is also an additional office in Germany, and there is a possibility to work with it during the analysis phase.

 

Management
Usually, they form a team of 5+ developers, technical lead and delivery manager. The client will have access to the whole team. Over the past 5 years, they failed to meet the deadline once, because of changes in client requirements.

 

Types of Engagement and Pricing
Models could be Fixed Cost or Time and Materials. Price varies, with the discount for a new client. You can make payments to the office in Germany.

 

Maintenance and Scaling
The company will help you with an evaluation of the scope change during the development. Depending on the case they define Service Level Agreement, but typically they provide 3 months warranty.

 

 

Advantages

  1. They have experience developing ML projects from scratch including Big Data expertise
  2. 4-5 hours overlap
  3. Access to the team
  4. Meeting the deadlines

 

Disadvantages

  1. Not flexible in the communication channels
  2. Poor information on pricing
  3. Lack of information on warranty

 

AI Company from Kyiv & Tel-Aviv – 4,6

Located in Ukraine and Israel they build and implement AI-powered systems across different verticals. COO replied in an impressive 30 minutes!

 

Expertise
Their team specializes in data science and data engineering, which means that most of their projects are already have a Machine Learning component. They presented three case studies with results: Vehicle Safety and Roadworthiness Test Predictor and Real-time Insurance Credit Score Modeling for automotive, as well as Computer Vision-Powered Personal Coach Platform for Amateur Athletes. Tech stack includes Python, Tensorflow, TensorflowLite, Flask, Keras, MongoDB, Docker, C++, Caffe, Mace, CoreML, Metal. They are also can give more information on case studies after signing NDA. This company has a University, and it already has over 1000 ML graduates.

 

Communications
Typically they have daily internal goal-setting and EOD reports by each team member working on a project. They communicate with the client twice a week, but it could be more often if necessary. Slack is the main communication service within a team. Communicating with client Whatsapp, Telegram or platform of your choice could be used. You will have access to CTO and COO all the time. All tasks are estimated and logged into Jira in real-time if you use Time & Materials model. They encourage the clients to visit their office at least once per quarter. Travel expenses of the development team are covered by the client with a separate bill, but it could be estimated before development as a part of the project.

 

Management
Their process starts with CTO understanding the project and what needs to be done. When architecture and technical stack are chosen they move to form a team. Usually, their team ranges from 3 to 9 people. That number may include Project Manager, Architect, 2 Data Scientists, 1-2 Data Engineers, DevOps, Front-End Developer, and Mobile Developer if needed. The entire team is located in Kyiv and will be accessible. They encourage you to visit the office when you want. To start the development you need to sign NDA, have a call to learn the specifications, agree on model/timeline and budget, sign the contract. Once approved, they start executing the timeline with specified stages. The first 3 stages take from 7 to 14 days.

 

They admit having situations in the past when the project took longer than expected. The reasons for that: delayed communication from the client, the client changed the list of features and added new ones to the MVP, the model took longer to train because of unexpected AWS infrastructure issues, data provided by the client wasn’t good enough. Budget-wise they have never billed the client more than was previously agreed on. All bill vs estimate argument was due to a feature request or the scope change, but in the end, there were no surprises for the client.

 

Types of Engagement and Pricing
There are three engagement models available:

  1. Time & Materials for long-term projects when the scope is open-ended.
  2. Fixed Cost for the cases when the scope is well-described
  3. Outstaffing for providing a dedicated team, recruiting/HR and office services for the client. The company is not responsible for the development team’s results because you will manage them.

 

Maintenance and Scaling
Their support typically includes overall maintenance, bug fixing, feature updates, and new feature releases. Data Scientist or Data Engineer is allocated part-time to support a project as long as needed. There was a case where they developed an AI solution in 6-9 months and provided support with senior data engineer for another 1.5 years.

 

 

Advantages

  1. The usage of the modern tech stack
  2. Experience with Big Data and High Load technologies
  3. Launching ML/DL models on mobile devices experience
  4. Reply from the COO in 30 minutes, the fastest among other companies in the article
  5. Flexibility in communication channels
  6. Access to the CEO and CTO
  7. Traveling to their office at least once a quarter
  8. Transparency and justified fails
  9. T&M, Fixed Price and Outstaffing models

 

Disadvantages

  1. Pricing was not mentioned
  2. It’s not clear whether maintenance is free or not

 

Conclusion

Reviewing the best AI companies we can say that they are a few things you must consider to make the right choice. Trustworthy Artificial Intelligence companies must prove their expertise and experience in this field, transparency in the prices and models, access to the development team, a variety of communication tools, openness in all processes and providing support periods. If you want to learn more about the companies in the article feel free to contact us, and we will provide more detail. We have done similar research on Top 6 Machine Learning Companies to Build Your AI Solution, you might be interested in it as well.