Martin Elias Costa
Verified Expert in Engineering
Martin is an experienced data scientist with vast knowledge of machine learning and artificial intelligence. After completing his Ph.D. in physics, he has worked on a wide range of projects from both the private and public sectors. He has extracted meaningful insights throughout the years and created products leveraging data sources in the most varied formats, including images, geospatial data, structured databases, time series, and text.
Visual Studio Code (VS Code), Jupyter, Python, Linux
The most amazing...
...thing I've developed is a deep learning image analysis pipeline for brain MRI images currently being used in a real life clinical setting.
Green Line, Inc.
- Scoped and designed all technical aspects of the project.
- Wrote the technical section for an NSF funding application.
- Researched the state of the art on all the key AI fields involved in the project, including large language models, text-to-speech, automatic speech recognition, and knowledge graphs.
The American Turmeric Company, Inc
- Developed an LLM-enabled chatbot using American Turmeric's custom data.
- Created a prototype UI for the chatbot and an API for easy integration with their existing site.
- Generated CI/CD code for deployment on Google Cloud Platform.
- Created update triggers to automatically track changes on a storage folder containing the knowledge source base documents.
NLP Engineer | Marketing Automation Startup
Channel Growth Inc
- Performed prompt engineering to improve automatic ad generation using GPT-3.
- Created a dynamic prompt template and database structure to generate ads tailored to each client's industry.
- Coordinated data collection and ran GPT-3 fine-tuning experiments.
Brightfield Group, LLC
- Designed and trained a new classification model for automatic product category extraction.
- Created custom reports and devised a customer segmentation strategy for better decision-making.
- Implemented an automated pipeline and dashboard for NLP-related KPI monitoring using Apache Airflow and Microsoft Power BI.
- Improved performance of existing product deduplication pipeline by 100 times.
- Developed several NLP models to automatically process students' notes. The tasks included: keyword-definition pairs extraction, automatic flashcard generation, and short answer grading.
- Suggested and designed new product features based on the latest NLP literature.
- Wrote technical requirements for externally hired software factories.
- Built an end-to-end ML pipeline for medical image analysis from the ground up. The system is currently serving clinics in Argentina, Brazil, and Chile.
- Trained deep neural networks for brain MRI segmentation and built automated reports to help radiologists make better decisions.
- Wrote production code to deploy the trained models in a real-life setting. Tested and reviewed code before deployment to cloud servers.
- Advised on key product decisions based on the latest literature for Medical Image Analysis.
National Government of Argentina
- Developed a flexible text-based internal search engine for the National Chief of Staff congressional sector.
- Created dozens of analytical reports on governmental datasets from different ministries.
- Developed computer vision algorithms to aid in the local monitoring of the mosquito Aedes aegypti, an important disease vector.
- Designed RF frequency controls to integrate into an existing system.
- Built the RF transmitters and incorporated them into the preexisting electronics.
- Deployed and tested the devices in a real-life setting.
- Evaluated the feasibility of a pilot test involving physiological measurements that may be indicative of emotional reactions of viewers while watching audiovisual material.
- Developed a prototype and gathered preliminary data.
- Wrote a detailed report with findings and recommendations.
Cazoll & Asoc.
- Designed a research protocol to evaluate skin moisture in users of cotton hygiene products.
- Oversaw a field study to obtain measurements from a large cohort of customers.
- Analyzed the data and wrote a detailed internal report summarizing findings.
National Survey Analysis for UNICEF
Research on Rat Communication
Educational Software Development
Data Science, Distributed Computing
Image Analysis, Natural Language Processing (NLP), Machine Learning, Artificial Intelligence (AI), Data Analytics, University Teaching, Data Analysis, Data Wrangling, Jupiter, Text Analytics, Supervised Machine Learning, Physics, Research, Statistics, Deep Learning, Medical Imaging, Computer Vision, Convolutional Neural Networks, Code Review, Source Code Review, Data Modeling, Image Processing, Data Engineering, OpenAI, Pattern Matching, Data Matching, Language Models, GPT, Generative Pre-trained Transformers (GPT), OpenAI GPT-3 API, Machine Learning Automation, ChatGPT, Large Language Models (LLMs), APIs, Data Visualization, Medical Diagnostics, OpenAI API, Neuroscience, Graph Theory, Statistical Modeling, Time Series Analysis, Geospatial Data, Electronics, Remote Sensing, Quality Assurance (QA), Object Detection, Technical Hiring, Interviewing, Mathematics, Robotic Process Automation (RPA), Ads, Leads, Marketing Automation, LangChain, Chatbots, Data Preprocessing, Feature Analysis, Cloud
Pandas, NumPy, Scikit-learn, jQuery, SpaCy, PyTorch, TensorFlow
Git, Jupyter, MATLAB, Apache Airflow, GIS
Linux, Amazon Web Services (AWS), Visual Studio Code (VS Code), Arduino, Google Cloud Platform (GCP), Docker, Amazon EC2
Amazon S3 (AWS S3), PostgreSQL
Ph.D. in Physics
University of Buenos Aires - Buenos Aires, Argentina
Master's Degree in Physics
University of Buenos Aires - Buenos Aires, Argentina