\documentclass[10pt,a4paper,ragged2e,withhyper]{altacv}
\geometry{left=1.25cm,right=1.25cm,top=1.5cm,bottom=1.5cm,columnsep=1.2cm} \usepackage{paracol}
\iftutex \setmainfont{Roboto Slab} \setsansfont{Lato} \renewcommand{\familydefault}{\sfdefault} \else \usepackage[rm]{roboto} \usepackage[defaultsans]{lato} \renewcommand{\familydefault}{\sfdefault} \fi
% Color scheme (Deep blue theme) \definecolor{SlateGrey}{HTML}{2E2E2E} \definecolor{LightGrey}{HTML}{666666} \definecolor{DeepBlue}{HTML}{0E3E68} \definecolor{AccentBlue}{HTML}{1567B1} \definecolor{LightBlue}{HTML}{D5E4F2}
\colorlet{name}{DeepBlue} \colorlet{tagline}{AccentBlue} \colorlet{heading}{DeepBlue} \colorlet{headingrule}{AccentBlue} \colorlet{subheading}{AccentBlue} \colorlet{accent}{AccentBlue} \colorlet{emphasis}{SlateGrey} \colorlet{body}{LightGrey}
\renewcommand{\namefont}{\Huge\rmfamily\bfseries} \renewcommand{\personalinfofont}{\footnotesize} \renewcommand{\cvsectionfont}{\LARGE\rmfamily\bfseries} \renewcommand{\cvsubsectionfont}{\large\bfseries}
\renewcommand{\cvItemMarker} \renewcommand{\cvRatingMarker}{\faCircle}
\begin{document} \name{Gusseppe Bravo} \tagline{AI Research Engineer | ML Systems Expert | PhD(c) in AI}
% \photoR{2.8cm}{photo.jpg}
\personalinfo{ \email{gusseppebravo@gmail.com} \homepage{scholar.google.com/citations?user=5JHTnBEAAAAJ} \linkedin{gusseppebravo} \github{gusseppe} \location{New York, USA · Open to relocation} \smallskip \quote{“If I can’t draw it, I can’t get it”} }
\makecvheader
\columnratio{0.6} \begin{paracol}{2}
\cvsection{Summary} \begin{itemize} \item \textbf{AI Research Engineer} at Barcelona Supercomputing Center with expertise in designing and deploying \textbf{production-grade AI systems}, specifically \textbf{AI-based agents} and \textbf{continual learning solutions}. \item \textbf{PhD candidate in AI} at UPC BarcelonaTech with publications in top-tier conferences (AAMAS, ICPR, CVPR) focusing on making AI systems more \textbf{robust}, \textbf{adaptive}, and \textbf{trustworthy} \item Delivered measurable business impact through AI innovation: \textbf{$200K in research funding}, \textbf{$100K in integration grants}, and \textbf{$10M+ in increased tax revenue} \item Specialized in bridging research-to-production gap with expertise in \textbf{MLOps}, \textbf{model monitoring}, and \textbf{AI system reliability} at scale \item Good at hacking skills (code, linux, problem solving). I’m not afraid of \textbf{thinking big}, at the end, we were able \textbf{to create everything} right? So for sure we can \textbf{solve hard problems}. \end{itemize}
\cvsection{Experience}
\cvevent{AI Research Engineer}{Barcelona Supercomputing Center \& Lenovo}{Apr 2020 – Present}{Spain/USA} \begin{itemize} \item Architected cognitive frameworks for Language Agents that \textbf{autonomously monitor ML model health}, reducing degradation incidents and improving model reliability \item Built TADIL algorithm for domain-agnostic continual learning, enabling \textbf{6\% performance improvement} across diverse domains without requiring task labels \item Developed Scanflow, an MLOps platform for debugging ML workflows, \textbf{some projects were built on top of it}. \item Developed a novel drift detection algorithm using autoencoder for drift detection on images. \textbf{Published in top-tier journal}. \item \textbf{Led cross-functional team}, resulting in several publications in top-tier journals and conferences \end{itemize}
\divider
\cvevent{AI Engineer Intern}{Emory Global Diabetes Research Center}{Jan 2024 – Jul 2024}{USA} \begin{itemize} \item Developed resource-efficient \textbf{cognitive architecture} for LLM-agents, including \textbf{finetuning small language models} like llama, phi for specific uses \textbf{reducing API costs by 35\%} while maintaining reasoning capabilities. Using small LMs and memories. \item Built an \textbf{algorithm} for healthcare \textbf{domain adaptation}, improving diabetic retinopathy detection \textbf{accuracy up to 3.1\%} on diverse patient populations (hard problem because of class-incremental learning) \item Published 3 papers on continual learning for medical imaging, with one selected as \textbf{spotlight} presentation at VISAPP conference \end{itemize}
\divider
\cvevent{AI Consultant}{SUNAT–Inter-American Development Bank}{Jan 2022 – Dec 2025}{Peru} \begin{itemize} \item Built the \textbf{first agent-based system} for customs to monitor ML models in production, enabling \textbf{real-time detection of customs fraud patterns} \item Built the first \textbf{“Improver” agent for automated ML model updates} that \textbf{reduced manual intervention by 65\%} while maintaining model performance \item Created SmartDeploy, the first end-to-end MLOps system for SUNAT, enabling \textbf{continuous delivery of ML models} for tax evasion detection \item Architected and deployed \textbf{enterprise-grade} infrastructure (MLFlow, JupyterHub, Minio, PostgreSQL, Ray) on Kubernetes % \item Developed novel drift detection library using autoencoder-based anomaly detection, \textbf{increasing model robustness by 42\%} in production \end{itemize}
% \cvevent{AI Consultant}{SUNAT–Inter-American Development Bank}{Nov 2024 – Dec 2025}{Peru} % \begin{itemize} % \item Pioneered the first agent-based system for customs to monitor ML models in production, enabling \textbf{real-time detection of customs fraud patterns} % \item Engineered an “Improver” agent for automated ML model updates that \textbf{reduced manual intervention by 65\%} while maintaining model performance % \item Implemented adaptive prompt architecture reducing token usage by 40\% while maintaining agent reasoning capabilities % \end{itemize}
% \divider
% \cvevent{Data Science Consultant}{SUNAT–Inter-American Development Bank}{Jan 2022 – Jan 2024}{Peru} % \begin{itemize} % \item Created SmartDeploy, the first end-to-end MLOps system for SUNAT, enabling \textbf{continuous delivery of ML models} for tax evasion detection % \item Architected and deployed enterprise-grade infrastructure (MLFlow, JupyterHub, Minio, PostgreSQL, Ray) on Kubernetes % \item Developed novel drift detection library using autoencoder-based anomaly detection, \textbf{increasing model robustness by 42\%} in production % \end{itemize}
\divider
\cvevent{Data Scientist}{SUNAT (Tax Administration)}{Nov 2017 – Nov 2018}{Peru} \begin{itemize} \item Engineered ensemble-based fraud classifier algorithm with \textbf{92\% accuracy} for identifying unjustified patrimony \item Developed graph analysis methods to uncover complex tax evasion networks, \textbf{processing 1TB+ of financial data} monthly \item Built scalable analytical infrastructure processing millions of electronic invoices, directly contributing to \textbf{$10M increase} in tax revenue \end{itemize}
\cvsection{Key Projects \& Open Source}
\cvevent{Cognitive Architecture for ML Monitoring}{github.com/gusseppe/cognitive_architecture_checker}{}{} \begin{itemize} \item LLM-based agent framework that \textbf{autonomously monitors ML models} in production environments \item Developed novel \textbf{decision procedure algorithm} and \textbf{structured memories} for interpretable model diagnosis \item Featured in \textbf{AAMAS 2025 conference}; developed in collaboration with Lenovo AI Lab \end{itemize}
\divider
\cvevent{Scanflow}{github.com/gusseppe/scanflow}{}{} \begin{itemize} \item \textbf{Production-grade MLOps} framework for ML workflow management, debugging, and reliability \item Many projects were inspired by this framework, like Scanflow-k8s, CAMA, KC-Agent. \item Published in top-tier journal “Expert Systems With Applications” (Impact Factor: 8.665) \end{itemize}
\switchcolumn
\cvsection{Core Competencies}
\cvskill{Production AI Systems}{5} \divider \cvskill{Python/Linux/MLOps}{5} \divider \cvskill{Continual Learning}{5} \divider \cvskill{Problem solving}{5} % \divider % \cvskill{AI Research \& Implementation}{5}
\medskip
\cvsection{Stack software} {\LaTeXraggedright \cvtag{Linux} \cvtag{Python} \cvtag{PyTorch} \cvtag{Numba} \cvtag{TensorFlow} \cvtag{LangGraph} \cvtag{AutoGen} \cvtag{Langchain} \cvtag{HuggingFace} \cvtag{MLflow} \cvtag{Ray} \cvtag{Docker} \cvtag{Kubernetes} \cvtag{RAG} \cvtag{Airflow} \cvtag{Spark} \cvtag{MongoDB} \cvtag{LLMs} \par}
% % \smallskip % {\LaTeXraggedright
% \par}
% % \smallskip % {\LaTeXraggedright % \cvtag{MLflow} % \cvtag{Ray/RayServe} % \cvtag{Docker} % \cvtag{Kubernetes} % \par}
% \smallskip % {\LaTeXraggedright % % \cvtag{Prometheus} % % \cvtag{Distributed Systems} % % \cvtag{Model Monitoring} % \cvtag{RAG} % \par}
\cvsection{Impact Metrics} \cvachievement{\faChartLine}{$250K+ Research Funding}{Secured from Lenovo and Intel for AI research initiatives}
\divider
\cvachievement{\faRocket}{$10M+ Revenue Impact}{Generated for Peruvian Tax Administration through AI fraud detection}
\divider
\cvachievement{\faCode}{40+ GitHub repos}{For open-source ML/MLOps projects and libraries}
\divider
\cvachievement{\faFile}{10+ Publications}{In top-tier AI journals and conferences (AAMAS, ICPR, CVPR)}
\cvsection{Education}
\cvevent{PhD in Artificial Intelligence}{UPC BarcelonaTech}{2021 – 2025}{} Research focus: Human-on-the-loop Continual Learning: Data, Knowledge and Agents for Model Adaptation \textbf{Dissertation submitted, defense pending}
\divider
\cvevent{MSc in Artificial Intelligence}{UPC BarcelonaTech}{2019 – 2021}{} Thesis: “Scanflow: A Learning-symbolic Framework for ML Workflow Debugging” \textbf{Published in Expert Systems With Applications (IF: 8.665)}
\divider
\cvevent{BSc in Computer Science, Magna Cum Laude}{National University of Engineering}{2011 – 2016}{} \textbf{Top 5\% of graduating class}
\cvsection{Awards}
\cvachievement{\faTrophy}{Intel-Lenovo AI Innovators Award (2023)}{Selected for innovation in continual learning and cognitive architectures research}
\divider
\cvachievement{\faTrophy}{Scanflow Integration Grant (2020)}{Project selected for integration into Lenovo’s enterprise AI platform}
\divider
\cvachievement{\faMedal}{María Rostworowski I Researcher (2020)}{National recognition for excellence in AI research}
\cvsection{Some publications}
\cvevent{Feature Engineering for Agents}{AAMAS 2025}{}{} % \textbf{Bravo-Rocca, G.,} et al. An Adaptive Cognitive Architecture for Interpretable ML Monitoring.
% \divider
\cvevent{Experience Replay and Zero-shot Clustering}{VISAPP 2025}{}{} % \textbf{Bravo-Rocca, G.,} et al. Continual Learning in Diabetic Retinopathy Detection.
% \divider
\cvevent{TADIL: Task-Agnostic Domain Incremental Learning}{ICPR 2024}{}{} % \textbf{Bravo-Rocca, G.,} et al. Task-ID Inference using Transformer Nearest-Centroid Embeddings.
% \divider
\cvevent{Scanflow: Multi-graph framework}{Expert Systems With Applications}{}{} % \textbf{Bravo-Rocca, G.,} et al. Machine Learning workflow management, supervision, and debugging.
\cvsection{Languages}
\cvskill{English}{5} \divider \cvskill{Spanish}{5}
\end{paracol} \end{document}