MARVIN-WEB-LOGO.png

Unlock the Potential of AI with Our Expertise

brain -mobile
  • Integrate advanced AI language model for enhanced customer interactions.
  • Automate queries and offer real-time assistance to customers.
  • Streamline operations, boost workflow efficiency, and drive engagement through personalized interactions.
  • Unlock AI’s full potential with a tailored custom-made engine
  • Maximize performance and efficiency with a bespoke AI solution
  • Empower your enterprise with a scalable cutting-edge AI engine
Schedule A Call

Schedule A Call

Skills & Experience

Skills

  • AI Integration Techniques
  • API Utilization
  • System Architecture
  • Natural Language Processing (NLP)
  • Conversational AI
  • Technical Stack: Python, ChatGPT API, NLP libraries
  • (such as NLTK or SpaCy), RESTful APIs
  • Data Science
  • Machine Learning
  • Deep Learning
  • Neural Network Training
  • Programming Languages: Python, TensorFlow, PyTorch
  • Technical Stack: Python, TensorFlow, PyTorch, NumPy,
  • SciPy, Pandas, Scikit-learn, CUDA, Neural Network
  • Architectures(CNN, RNN, GAN, etc.)

AI Techniques

  • Prediction (Supervised)
  • CV Algos
  • Prediction (Hebbian learning), Decision making AI module
  • Multi Layer Perceptron
  • Unsupervised learning
  • Object detection, classification
  • computer vision and statistics
  • Random Forest (Supervised)
  • Path planning
  • Synthetic data
  • Template matching
  • Bundle adjustment and SLAM
  • Reinforcement learning

Open source projects:

-CV-based segmentation, detection

  • Langchain: NLP
  • Segment anything: segmentation computer vision
  • Mlflow: mlops

Relevant APIs

-AWS SageMaker

-GCP AI

-Azure ML

-OpenAI

Projects

Robotic Navigation cover

Robotic Navigation

Medical Spine Fix (Scoliosis)

Customers Segmentation

Robotic Process Automation

Robotic Process Automation

Social Media analysis

Social Media analysis

Evalmee

Evalmee

City of Paris

City of Paris

La Brasserie Barge

Plate numbering recognition

Plate numbering recognition

Contact Us

cropped-MARVIN-WEB-LOGO.png

We are ready for the challenge!

Info@marvin.live

Linkedin

The Challenge:

  • Planning the optimized path of
    multiple robots in a warehouse
  • Collusion detection
  • Robot motion smoothing

Solution :

  • Mathematical motion
    smoothing for turns and
    movement
  • Optimized route using state of
    the art algorithms

Business Value :

  • Outsource of a complete problem.
  • Module ready to use
  • Ability to focus on core areas of the company
  • AI Techniques: Path planning

The Challenge:

  • Optimize a medical device
    location
  • Computerize Doctors observation
    of the scoliosis
  • Vertebrae deformation detection

Solution :

  • Complete diagnosis of scoliosis
  • AI-guided medical instrument
    location

Business Value :

  • More accurate measurements than human Doctor
  • FDA Ready development process
  • Leverage the medical procedure method to WW physicians
  • AI Model reconstructing is critical for FDA approval

The Challenge:

  • Improve sales and conversions
  • Campaign Optimization
  • Identify failure points
  • Market fit – Identify demand at
    product discovery phase
  • Multiple Data Sources

Solution :

  • Automatic Clustering
  • What their clients need?
  • Why they need it?

Business Value :

  • New demand groups discovered
  • – were under the radar!
  • Funnel Increase of 50%
  • AI Techniques: Prediction (Supervised)

The Challenge:

  • Limited code installation on a
    remote device.
  • Detecting type of forms and
    construct into automated
    process
  • Small data reference

Solution :

  • 100% accuracy

Business Value :

  • Savings on manpower
  • Client sorted their valued data.
  • AI Techniques: Synthetic data, Template matching, Deep Learning

The Challenge:

  • Success Metrics and Dashboard
  •  Content recommendation
  •  Increase engagement

Solution :

Rebuild data flow
Data warehouse
KPI and metrics dashboard
Continuous AI with various models
on an MLOps platform.

Business Value :

  • 50% increase in engagement after 3 months of Data-Centric system operating
  • Getting the engines and results of the big social network players within months
  • AI Techniques: Prediction (Supervised), unsupervised learning

Context :

Evalmee is a startup that automatically oversees students during their examinations. They use a camera to assess whether a student is cheating or not.

Mission :

We designed machine-learning-based algorithms to assess the probability that a customer is cheating. To scale with millions of frames, we used an ensemble of very lightweight deep-learning-based models. We build a dataset to evaluate our algorithms. We deploy our solutions on AWS.

Skills :

  • machine learning algorithms
  • computer vision
  • annotations
  • cloud computing

Context:

The city of Paris conducts each year participatory budgeting in which inhabitants decide how to allocate part of the municipal budget. The city has roughly 100k paper ballots and 50k electronic ballots.

Mission :

We worked closely with the organization to upgrade their application, using an innovative ballot system, and enhance the overall user experience. We automatically decoded the paper ballots using advanced computer vision-based algorithms.

Skills :

  • machine learning algorithms
  • computer vision
  • user experience
  • web-developpement

Context:

La Brasserie Barge is a brewery in the Greater Paris. It sels beerr to supermarkets, bars, direct customers, and festivals.

Mission :

We have consolidated their customer databses, provided data visualization, customer lifetime prediction, and a marketing roadmap to allow them to scale up. We have also automatized several processes in the customer acqusition funnel.

Skills :

  • data-science
  • data-visualization
  • automatization tools

Context :

Stanley Robotics is a startup that builds a robot lifting and 

moving cars autonomously. Customers leave their vehicles in a 

box and the robot takes care of it.

Mission :

On a box, 5 cameras are picturing the license plate attached to a 

car with high distortion. We designed machine-learning-based 

algorithms to recognize the plate’s numbers. We build a dataset 

with high-level annotations. We trained and evaluated our 

algorithms on this dataset. 

Finally, the models have been using cloud computing resources 

to serve thousands of customers per day.

Skills :

  • machine learning algorithms
  • computer vision
  • annotations
  • cloud computing
Managed Customized Development Services