Jose Marcos Rodriguez Fernandez, Developer in Barcelona, Spain
Jose is available for hire
Hire Jose

Jose Marcos Rodriguez Fernandez

Verified Expert  in Engineering

Machine Learning Developer

Location
Barcelona, Spain
Toptal Member Since
August 7, 2019

Marcos is a computer scientist with a master's degree in artificial intelligence. He has 7+ years of experience in development and research yielding enterprise-ready artificial intelligence (AI) and machine learning (ML) projects in the areas of computer vision (CV) and natural language understanding (NLU). He's also co-founded an AI company developing conversational AI technology and managed large remote tech teams in different time-zones.

NLUNatural Language Processing (NLP)Machine LearningComputer VisionDeep Neural NetworksDeep LearningPythonScikit-learnPyTorchUbuntuJSONNatural Language Toolkit (NLTK)GitHubUnit TestingTensorFlowSupport Vector Machine

Portfolio

Melior.AI
Amazon Web Services (AWS), Cloud, Elasticsearch, MongoDB...
DigitalGenius Ltd
Spark, MySQL, TensorFlow, PyTorch, Torch, Python
Serimag S.L
OpenCV, Neural Networks, Support Vector Machines (SVM), Scikit-learn, C++...

Experience

Python - 8 yearsComputer Vision - 7 yearsMachine Learning - 7 yearsNatural Language Processing (NLP) - 7 yearsGPT - 7 yearsGenerative Pre-trained Transformers (GPT) - 7 yearsTensorFlow - 3 yearsNoSQL - 3 years

Availability

Part-time

Preferred Environment

GitHub, Visual Studio Code (VS Code), Ubuntu

The most amazing...

...project I've worked on is an AI-based language agnostic dialogue system capable of integrating text, voice, image, and other events.

Work Experience

Co-foudner and CTO

2018 - PRESENT
Melior.AI
  • Co-founded Melior.AI.
  • Researched and developed novel technologies in the space of NLU and dialogue understanding.
  • Developed and evaluated NLU technology for chat outperforming several of the major cloud providers and platforms at the day of publishing: https://medium.com/melior-ai/natural-language-understanding-part-2-83c27fe651cc.
  • Defined technology strategy and road maps.
  • Defined budgeting and hiring strategy.
  • Pitched in front of investors and clients.
  • Presented in conferences and incubators.
  • Managed technological development with partners and clients.
  • Managed fully remote teams of up to 12 engineers across different time-zones in several fields of technology.
Technologies: Amazon Web Services (AWS), Cloud, Elasticsearch, MongoDB, Convolutional Neural Networks (CNN), Neural Networks, NLU, Python

Deep Learning Research Engineer

2015 - 2018
DigitalGenius Ltd
  • Created language-agnostic natural language technology based on deep learning for question-answering systems.
  • Researched and developed the first prototypes of state of the art Q&A deep learning-based solutions in open datasets and benchmarks. Outperforming IBM Watson in the insurance QA dataset.
  • Co-authored and published research in NLU: "An Attention Mechanism for Neural Answer Selection Using a Combined Global and Local View": https://arxiv.org/pdf/1707.01378.pdf.
  • Co-authored several machine learning-based algorithms for NLU resulting in patent submissions both in the EU and USA.
  • Developed several optimization algorithms for multi-tenancy deployments of a machine learning model to enable on-demand real-time model loading in memory.
  • Developed a new attention mechanism for neural-based question answering, which depends on varying granularities of the input.
Technologies: Spark, MySQL, TensorFlow, PyTorch, Torch, Python

Machine Learning Engineer

2013 - 2015
Serimag S.L
  • Developed automatic document processing technology based on machine learning approaches for classification and information extraction, mainly (but not only) for bank documentation for CaixaBank.
  • Developed a multi-queue and multi-processing processing system for high throughput real-time machine learning-based extraction and verification software for bank checks.
  • Used computer vision techniques to correct, identify, and extract different fields in noisy multi-page documents (import, export, invoices).
  • Developed OCR and semantic feature extraction solutions for image-based documents from scans.
  • Developed fast text processing algorithms in C++ and its python wrappers to serve as common text-processing routines for all major projects.
Technologies: OpenCV, Neural Networks, Support Vector Machines (SVM), Scikit-learn, C++, Python

HOG Pedestrian Detector

https://github.com/jmrf/HOG-Pedestrian-Detector
This work targets pedestrian detection in static images from a computer vision point of view. The interest of such detector resides in its many applications; automotive safety, crowd control, video surveillance or automatic image indexing are just a few examples. Detecting pedestrians is a challenging matter as persons can adopt a wide range of poses, in very different backgrounds and under significant changes in illumination and color. To achieve a robust detection method we study and develop a HOG plus SVM solution, as proposed by Dalal & Triggs. The HOG descriptor proposed turns out to be robust to small changes in the image contour, location and direction, and significant changes in illumination and color. Even though HOGs perform equally well for other classes, in this work we target specifically in upright persons, or to say, pedestrians. Furthermore, we try several SVM models and training approaches to pick out the best possible SVM kernel and parameters for two different well-known person data sets; MIT and INRIA data sets
2014 - 2015

Master's Degree in Artificial Intelligence

FIB - UPC - Barcelona, Spain

2008 - 2013

Master's Degree in Computer Science

FIB - UPC - Barcelona, Spain

JANUARY 2017 - PRESENT

Probabilistic Graphical Models

Stanford University

JANUARY 2016 - PRESENT

Object Detection

Universitat Autònoma de Barcelona

JANUARY 2015 - PRESENT

Neural Networks for Machine Learning

University of Toronto

JANUARY 2014 - PRESENT

Machine Learning

Stanford University

Libraries/APIs

PyTorch, Scikit-learn, TensorFlow, Flask-RESTful, Natural Language Toolkit (NLTK), OpenCV, Node.js

Tools

MATLAB, Scikit-image, GitHub, GitLab CI/CD, Postman

Languages

Python, Prolog, C++, Java, JavaScript

Paradigms

Unit Testing, Kanban

Frameworks

Spark, Flask

Platforms

Ubuntu, Visual Studio Code (VS Code), Amazon Web Services (AWS)

Storage

MongoDB, NoSQL, JSON, MySQL, Elasticsearch, Redis

Other

Remote Work, NLU, Natural Language Understanding (NLU), Natural Language Processing (NLP), Deep Neural Networks, Deep Learning, Machine Learning, Computer Vision, GPT, Generative Pre-trained Transformers (GPT), Support Vector Machines (SVM), Feature Roadmaps, APIs, Version Control, Neural Networks, Torch, Convolutional Neural Networks (CNN), Cloud

Collaboration That Works

How to Work with Toptal

Toptal matches you directly with global industry experts from our network in hours—not weeks or months.

1

Share your needs

Discuss your requirements and refine your scope in a call with a Toptal domain expert.
2

Choose your talent

Get a short list of expertly matched talent within 24 hours to review, interview, and choose from.
3

Start your risk-free talent trial

Work with your chosen talent on a trial basis for up to two weeks. Pay only if you decide to hire them.

Top talent is in high demand.

Start hiring