Victor Maricato Senior Machine Learning Engineer @ Upwork

Senior Machine Learning Engineer

Upwork

Sep 2022 - Current (3 years 8 months)

Part of Upwork's Algorithms & Research team. Mainly working on LLMs, RAG, knowledge graphs and graph neural networks. * Led the ML collaboration with the product team to create a rule-based metric on the team's performance, allowing them to create baselines and optimize their work. * Migrated complex graph-based ML models from Databricks to AWS, building and optimizing reusable features for a Feature Warehouse. This ML initiative significantly helped cut costs and increase the company's revenue. * Led team of Machine Learning engineers on Databricks to AWS SageMaker model migration. * Technically leading development, productionization, and quantitative tests for a new deep-learning-based search engine in the platform. * I am on the cen

M

Research Assistant

MIT Computer Science and Artificial Intelligence Laboratory (CSAIL)

Sep 2021 - Sep 2022 (1 year 1 month)

I worked on protein language model research with large language models (LLM) and retrieval-augmented generation (RAG). Personally responsible for implementing the research code and conducting experiments. This is one of the first studies to apply the RAG technique to power protein language models in a multi-modal architecture. Advisor: Luis Alves

Genomic Data Scientist

Mendelics

Sep 2020 - Mar 2022 (1 year 7 months)

Collaborated with seven colleagues (with multidisciplinary backgrounds such as biotechnology, medicine, biomedicine, etc.) for developing state-of-the-art products as a research-focused team. * Responsible for developing a machine learning model focused on variant predictions (pathogenicity prediction). This model required the use of unstructured data (DNA, RNA, Protein sequences) with a large number of tokens. The final model beat the state-of-the-art Illumina's model (PrimateAI) in detecting pathogenic variants, increasing sensitivity in more than 10% and achieving almost twice the specificity. * Personally responsible for deploying the machine learning product that identified SARS-CoV-2 variants in Brazil. This was done as an emergency

Data Scientist

OLX Brasil

Apr 2019 - Sep 2020 (1 year 6 months)

In the central Data Science team at OLX I and my colleagues have managed to: * Churn prediction using sequence models. We proved through A/B Testing that the model could detect with more than 90% sensitivity which clients would churn within 2 weeks. This information was then sent to the Customer Relationship teams in order to increase customer retention. Tableau was used to track the experiment results. * Computer Vision model to suggest item category. Applied transfer learning on a previously existing image model as a feature extractor and then fine-tuned the downstream model to classify a product item into a pre-existent category. This project also demanded Android Development skills in order to implement tracking events to measure the

Biotechnology Data Scientist

Shell

Jul 2018 - Mar 2019 (9 months)

Worked on bioproducts yield and morphology predictive systems and optimization. Increasing bioproduct yield in 30% through Box-Benkhen, Genetic Algorithm, and Deep Learning optimization models. Several challenges in working with Design of Experiments for Machine Learning algorithms in small data environment.

Biotechnologist

Shell

Dec 2017 - Mar 2019 (1 year 4 months)

Biopolymers development for Pre-salt enhanced oil recovery. Applied biochemistry, microbiology, and bioinformatics knowledge. Coordinators: Alane Beatriz Vermelho (BioInovar - IMPG/UFRJ); Claudia Elias Regina Mansur (IMA/UFRJ)

Research Internship - Cell Biology, Biochemistry and Cancer Research

Federal University of Rio de Janeiro

Apr 2017 - Nov 2017 (8 months)

Laboratory: Laboratory of Biochemistry and Cell Biology of Glycoconjugates (LBBCGlico) Worked on the project: "Extracellular Galectin-3 interactions with sulfated polysaccharides" Advisors: Mauro S G Pavão; Felipe C O B Teixeira; Mariana P Stelling

Research Internship - Bioinformatics

Universidade Federal do Rio de Janeiro

Nov 2016 - Jan 2017 (3 months)

In the Laboratory of Bioinformatics (R. 535), I was responsible for migrating a local script in Ruby to the cloud bioinformatics platform named "Galaxy".

Research Internship - Inflammation, Nanotechnology, Cancer Research

FIOCRUZ - Fundação Oswaldo Cruz

Aug 2016 - Nov 2016 (4 months)

In this research, I and my colleagues tested Pequi Oil for Alzheimer's and Astrocytoma treatment.

Research Internship - Plant Biology

IFRJ Instituto Federal do Rio de Janeiro

Feb 2016 - Nov 2017 (1 year 10 months)

In this project, I learned and aid in the research's statistical analyses. * Survival analyses indicated that after the disaster, planting on the affected soil would result in significantly less yield. * Liquid Chromatography indicated that the cause for the reduced soil fertility may have been caused due to heavy metal and Iron accumulation.

Victor Maricato

About

Detail

Timeline

Résumé