Build complete AI application by Commercial off-the-shelf AI services
Leverage the modern AI frameworks to build the AI solutions for the business to boost performance and innovation from law firm, education, BPS, to E-commerce.
Generative AI - Limitless AI Applications
Using deep learning algorithms and large models to train for billions of hyper-parameters to create new digital content such as images, video, audio, text or code.
Background image: A long journey on a gravel path of Mars - generated by AI
Generative AI capabilities
Text, code generation
AI can rephrase a paragraph to correct your input or summarize articles or news. It can also generate code for developers to boost performance.
Image, video generation
AI can generate the images like artists or can generate the video for based on the context. (e.g., deepfake technology)
Audio generation
Build text-to-speech AI voice generator with your own voice. You can train AI by inputting your voice samples.
Want to explore Generative AI capability at NashTech?
Benefits from AI solution accelerator
Building a full AI solution is tough and takes time. With AI solution accelerator template, you can quickly create an end-to-end AI project skeleton with lot features in place such as Nvidia-docker configured, microservices and DevOps ready. You can now focus on AI feature engineering, model building & evaluation.
Quickly building AI projects based on AI solution accelerator template.
Reusable libraries are collected and ready to apply at production.
Transparency on what algorithms have been used, explainable by documents and can be extended with provided source code.
NashTech AI engineers are familiar with the libraries and can boost the performance if using the AI Lib.
Features to fulfill all smart AI applications
Full end-to-end AI solution
With couple of steps using the CLI, we will get the full end-to-end AI template ready to run. The template includes: a simple AI module, REST API, Nvidia-Docker and ready to run on K8S or on the cloud.
Modern architecture
The template uses microservices architecture which allow develop multiple AI models with ease. The AI model pipelines communicate via ZeroMQ and it's scalable or scale to zero using KEDA.
MLOps
Maintaining the life cycle of ML models is very important. The template uses MLflow to evaluate, publish and deploy the models across the environments. It also include set of Terraform code to enable IaC as well as using Azure DevOps to automatically build & deploy new AI modules.
Cloud ready
The template is built with cloud-native approach in mind. We use K8S, Helm chart, Nvidia-Docker to bundle the AI modules into Docker images and easily deploy to popular cloud providers (Azure, AWS, GCP). Also, we can deploy the template to Azure ML which the template is compatible with.
Ready-to-use AI modules
Going along with the AI template, NashTech has built lot of ready-to-use AI modules such as Face recognition, Form processing, Q&A, etc. We leverage open-source the state-of-the-art models and package them into the AI modules in our template. It's FREE and you can use it for commercial purposes.
Modules customized to fit your business
We have built customized AI app based on the great AI models from the giant companies such as Google, Facebook, Microsoft & Nvidia. The list below are our libraries to help client usage for free.
Faces recognition
Recognize face in the images, ages and expression prediction
MIT LICENSE
Form processing
Extract data from PDF and image files
MIT LICENSE
Question & Answer
AI learns the context and answers the questions
MIT LICENSE
OCR
Extract text from PDF or image files
MIT LICENSE
Keyboard dynamic
Detect identity based on keyboard dynamic patterns
MIT LICENSE
Card reader
Extract data from ID card, visit card and driver license
MIT LICENSE
Sentiment analysis
Recognize the sentiment in a paragraph
MIT LICENSE
Resume parser
Extract skills, profile info from PDF resume
MIT LICENSE
Cloth tryon
Try cloth to see it fit your body
COMMERCIAL LICENSE
Technologies behind the modern AI architecture
NashTech has researched on modern AI toolset such as Kubeflow, MLflow or cloud version such as Azure Machine Learning, AWS SageMaker to build modern AI applications. Also, We have built our own architecture which to adapt well with existing modern architecture such as microservices, containerization.
Supported cloud platforms
Built-in technologies