Leandro M. Abraham
leandromaf@gmail.com
linkedin.com/in/leandromaf
Digital CV
About me
Expert in Artificial Intelligence, Data Science, Machine Learning, and Software Engineering with theoretical knowledge and more than 10 years of experience in Research & Development and applied projects.
Recently specializing in Causality, Deep Learning, and Large Models (Language, Vision, or Multimodal) with great interest in applications in Medicine and Biosciences.
Academics
AI for Medicine
Coursera - June 16th, 2025 - License: 2PZ1EWT8N3CX
Computer Science PhD
Universidad Nacional del Centro de la Provincia de Buenos Aires, Facultad de Ciencias Exactas, Tandil - October 26th 2018 -
Thesis: Computer Vision and Machine Learning applied to biceps muscle activation estimation
Deep Learning Specialization
Coursera - March 18th, 2018 - License: DK2MNP5NQSH4
Technological Management Specialization
Universidad Nacional de Cuyo, Facultad de Ciencias Económicas, Mendoza - August of 2017
Final Work: R&D&i Agile Management and LEAN StartUp method to agile building of innovative StartUps
Universidad Tecnológica Nacional, Facultad Regional Mendoza - December 19th 2011
Professional experience
- Technical Interviewing
- Engineers mentoring in Data Science, Machine Learning, and Data Processing
Senior Artificial Intelligence Engineer for TripleLift (January 2022 - Now)
- Research, Experiment, and adopt Artificial Intelligence technologies to develop Intelligent Systems and Services.
- Design and Analysis of metrics to evaluate a Product feature success
- Collaboration with Product Managers, Cloud, and MLOps Engineers in the planning and architecture design to develop both Proof of Concepts as well as Production Intelligent solutions for Video and Image analysis using Large Language and Vision models and classical Computer Vision and Machine Learning techniques
- Design and Development of a RAG service to Categorize Web Ads following the IAB Ad Product Taxonomy using Large Multimodal Models and Embeddings
- Lead communications with the Product Manager to define and handle requisites and expectations
- Designed and implemented the main parts of the system
- Experimentation with different embedding models ending up using AWS Titan Multimodal Embeddings model
- Collaboration with Backend engineers to use this endpoint in the product and evaluate the predictions in production
- Design and Development of a Computer Vision system to detect digital overlaid graphics and text on the Screen
- Lead and design data labeling process through AWS Sagemaker Ground Truth
- Model training and optimization on Sagemaker Notebooks and Training jobs
- Deployment as server-less endpoint on AWS Sagemaker infrastructure
- Collaboration with Backend engineers to use this endpoint in the product
- Collaboration in the implementation of a dataset manager and evaluation system to quickly experiment and iterate different services based on MySQL databases, and Python scripts activated through Github actions and services implemented as lambda and step functions in AWS architecture
- Agile and fast state-of-the-art research, prototyping, and implementation of Proof of Concept pipelines for various potential CTV Content analysis types like:
- Emotion recognition from speech (using a HuggingFace implementation of a model pre-trained on the GoEmotions dataset)
- Audio tagging (using a model pre-trained on the AudioSet Ontology)
- Characters analysis (using the DeepFace Python library)
- Design and Development of a Computer Vision system to cluster images and shots as being part of the same scene
- Problem and dataset definition, metrics proposal, state-of-the-art review, and solution proposal
- Lead and design data labeling process through AWS Sagemaker Ground Truth
- State-of-the art Vision Transformers (ViT) models experimentation and customization
- Implementation of a pipeline for model evaluation and version comparison
- Design and Development of a Computer Vision system Proof of Concept to re-identify all instances of a query object in a video
- Problem and dataset definition, metrics proposal, state-of-the-art review, and solution proposal
- Benchamark evaluation dataset building from multiple sources
- State-of-the-art models experimentation and customization
- Implementation of a Proof of concept inference pipeline for model evaluation
- Collaborate in the implementation, deployment and evaluation of Machine Learning services based on pre-trained and open-source Large Language and Multi modal models used to understand, describe, and perform zero-shot classification on images and videos.
- Experimentation and evaluation with the following models (among others) :
- Collaborate in the design and implementation of an evaluation pipeline to evaluate the aforementioned models
Senior Data Scientist (August 2020 - January 2022)
Data Analysis, design, and experimentation of Predictive Models using Python for:
- Development and deployment of Personalized events recommendation models
- Event pages impressions prediction and analysis
- Multi-touch marketing channel attribution
- Creator profile page engagement prediction
- Event creators segmentation
- Junior Engineers mentoring in Data Science, Machine Learning, Experimentation and Data Instrumentation
Data Scientist (part-time, October 2019 - July 2020)
Data Analysis, design, and experimentation of Predictive Models using Python for:
- Product features usage understanding
- Tickets re-seller detection
- External Apps recommendation for event-creators
- Event attendance prediction. Collaborated to develop a model that predicted the fraction of event attendees from the ticket buyers within an 8% margin of error.
Back-end Software engineer (January 2019 - July 2020)
Collaborating with the design and implementation of services using Python, for:
- Real-time tracking of product usage
- Serving Machine Learning models for in-product insights and recommendations for Event Creators. Price recommendation model to help creators set the ticket prices for their events. Product feature that resulted in a conversion rate of > 90% with a lift of > 750K in gross revenue for creators and more than 80% positive NPS scores
- Bulk event information retrieval for external partners
Scrum Master (January 2019 - October 2019)
Neoproteins (August 2020 - December 2021)
Artificial Intelligence Adviser in R&D to find non-animal replacements for animal food proteins.
- Proteins Database processing and analysis
- Machine Learning Model Development and evaluation to predict protein function (e.g.: Solubility)
- Collaborate to develop PoC, business plans and product ideation
Dymaxion Labs (December 2018 - July 2020)
Team coordination and technical consultancy in image analysis and knowledge extraction with Computer Vision and Machine Learning. Mainly working with Deep Learning and overhead image processing Python libraries for:
- Meteorological time series prediction
- Land use classification
- Pavement images analysis and crack detection
- Roofs and buildings detection
- Informal settlements detection
Family Guard (December 2016 - December 2019)
Design and implementation of Back-end processes and pipelines that use Machine learning Predictors for GPS data analysis to detect anomalies in daily routines and potential danger.
- Design and implementation of a back-end for Natural images digit detection and recognition for credit card OCR
- Developing a proof of concept EEG time series predictor to simulate possible outcomes of epilepsy surgery with Recurrent Neural Networks
- Implementing an NLP engine for classification and tagging of legal documents accessible through a REST web service.
Teaching & Research
Artifical intelligence researcher at DHARMa Research Laboratory (since March 2011)
Applied scientific research in :
- RGB image and 3D point cloud classification
- Research paper Writing
- R&D and Data Science consultancy services
Teacher - Object Oriented Design and Programming
Universidad Tecnológica Nacional (Facultad Regional Mendoza) - October 2018 to July 2019
Teaching assistant - Artificial Intelligence
Universidad Tecnológica Nacional (Facultad Regional Mendoza) - Information Systems department - from 2013 to 2023
Teaching assistant - Machine Learning
Universidad Tecnológica Nacional (Facultad Regional Mendoza) - Information Systems department - 2012, 2013, 2017 and 2018
Technologies
- Python (
pandas, seaborn, numpy, scikit-learn, pytorch, torchvision, keras, flask, django )
- AWS Sagemaker, S3 and EC2
- SQL
- R
Postgraduate courses & trainings
AI For Medical Treatment
Coursera - License: 816DV5UEZF8V - July 2025.
AI for Medical Prognosis
Coursera - License: 816DV5UEZF8V - August 2024.
Practical Data Science with Amazon SageMaker
AWS - Certificate - December 11, 2023
AI for Medical Diagnosis
Coursera - License: BMFWHX6MW6TF - October 2023.
MLOps Engineering
AWS - Certificate - December 21, 2022
Machine Learning Explainability
Kaggle - April 2020
Sequence Models
Coursera - License: NKXGEC2RJQ8K - March 18, 2018
Convolutional Neural Networks
Coursera - License: WV9TKJT5HZXE - February 25, 2018
Structuring Machine Learning Projects
Coursera - License: TMNAMU5QX5PG - January 6, 2018
Improving Deep Neural Networks: Hyperparameter tuning, Regularization and Optimization
Coursera - License: SFGHHVEYQC9W - December 22, 2017
Neural Networks and Deep Learning
Coursera - License: X3HCBKEYHUSU - December 2, 2017
Computer Vision
Universidad Nacional del Sur - Bahia Blanca, Buenos Aires, Argentina - September 2013
Scientific Visualization
Universidad Nacional del Sur - Bahia Blanca, Buenos Aires, Argentina - September 2013
Introduction to planning algorithms
Universidad Nacional del Centro de la Provincia de Buenos Aires - Tandil, Buenos Aires, Argentina - August and September 2013
Introduction to evolutionary computing
Universidad Nacional del Centro de la Provincia de Buenos Aires - Tandil, Buenos Aires, Argentina - June and July 2013
Universidad Nacional del Centro de la Provincia de Buenos Aires - Tandil, Buenos Aires, Argentina - May 2013
Web Data Mining
Universidad Nacional del Centro de la Provincia de Buenos Aires - Tandil, Buenos Aires, Argentina - May 2013
Digital images computational analysis
Universidad Tecnológica Nacional - Facultad Regional Mendoza - Mendoza, Argentina - From April 2012 to June 2012
Publications
Abraham L. and Bromberg F. - Visión computacional y aprendizaje de máquinas aplicado a la estimación de activación muscular del bíceps braquial
PhD thesis in Computer Science. Universidad Nacional del Centro de la Provincia de Buenos Aires. October 2018.
Abraham L., Bromberg F. and Forradellas R. - Ensemble of shape functions and support vector machines for the estimation of discrete arm muscle activation from external biceps 3D point clouds
Computers in Biology and Medicine, Volume 95, 2018, Pages 129-139, ISSN 0010-4825,DOI.
Abraham L., Bromberg F. and Forradellas R. - Arm muscular effort estimation from images using Computer Vision and Machine Learning
Proceedings 1st International Conference on Ambient Intelligence for Health, Puerto Varas (Chile), December 2015, LNCS 9456 (ISBN: 978-3-319-26507-0)
ASAI 2014, Buenos Aires (Argentina), September 2014.
Schlüter F., Bromberg F. and Abraham L. - Strategies for piecing-together Local-to-Global Markov networks learning algorithms
ASAI 2011, Córdoba (Argentina), August 2011.
Languages
Spanish
Native
English
Proficient conversation, writing, and reading
EF SET C1 Advanced Certified