Bruna Lemberck

Owned the development of the company’s next-generation Generative AI products, focused on automating and enhancing user journeys across Search & Match and the Talent Acquisition Platform. - End-to-End GenAI Delivery: Designed and implemented GenAI solutions across the full lifecycle—from PoC and MVP to production-scale systems. - Multi-Agent Candidate Search & Matching: Built multiple multi-agent systems powering intelligent candidate search and job matching, including: 1. Algolia + LLM RAG Pipelines: Architected LLM pipelines leveraging Algolia to retrieve and match registered candidates (including CVs) against open job opportunities with high relevance and fast response times. 2. Skill & Profession Extraction: Developed LLM-based extraction and normalization pipelines to structure skills/professions from unstructured text using taxonomy-driven mapping for normalization with similiarity search. 3. LLM-as-a-Judge Evaluation: Implemented LLM-as-a-Judge evaluation workflows to validate quality, relevance, and accuracy of matching outcomes. - LLMOps & Observability: Integrated Langfuse for monitoring, tracing, and continuous improvement of LLM workflows. - Data Engineering & BI Data Analysis: Built and maintained batch and near-real-time pipelines to support reliable analytics and visibility into business and product metrics; produced executive-ready insights through ongoing Python-based analysis for strategic decision-making. >> Stack: Google Gemini, Vertex AI, Vercel AI, OpenSearch, AWS (Lambda, DynamoDB, SQS, CloudWatch, OpenSearch), GCP, Langfuse (LLMOps), Cursor AI, Claude Code, Node.js (TypeScript/JavaScript), RAG, similarity search, multi-agent systems, Docker, Algolia.

Led Generative AI projects for startups and consultancies, project-based, from gathering client requirements to designing and implementing solutions. Built MVPs and PoCs to turn ideas into working prototypes. Mentored data and software teams on GenAI, guiding practical applications and creating learning paths to empower them in developing AI solutions. _______________ || Key Projects || part 1 - SQL Agent: Designed and developed an MVP of a ReAct SQL agent using a multi-agent architecture for Natural Language Query (NLQ), allowing non-technical users to translate their questions into SQL queries tailored to the company's database structure. The agent automatically discovers relevant tables and schemas based on user queries, making data access more intuitive. >> Stack: LangGraph, Langchain, Langsmith, OpenAI, Python - LLM Routing: Developed a strategy to optimize system costs and improve reliability by dynamically selecting the best AI model for each task of a customer service multiagent system. >> Stack: OpenRouter, Langchain, LangGraph, OpenAI (GPT-4o-mini, GPT-4o), Anthropic (Claude Sonnet), Mistral AI (7b) - Advanced RAG System: Built a PoC for a RAG system using advanced techniques (C-RAG, Self-RAG, Adaptive-RAG). The system retrieves information from a vector database or performs a Google search based on user queries, providing precise and context-aware responses, and uses 2 layers of self-rag to check for hallucinations and relevance, ensuring the response directly answers the user's question. >> Stack: LangGraph, Langchain, Langsmith, OpenAI, Python - Multimodal Embedding: Developed a PoC for a RAG system with support for multimodal embeddings, enabling it to understand and process PDF tables, images, and hierarchical text structures, to interpret complex documents and answer user queries accurately. >> Stack: JinaAI (multimodal embedding, CLIP model), Unstructured API (chunking strategy), Qdrant (VecDB), Langchain, Langsmith, OpenAI, Python

_______________ || Key Projects || part 2 - Multi agents System for full Lead Scoring and classification : Developed the MVP of a multi-agent system chatbot with RAG. This chatbot interacts with users, introduces the company, and captures basic user and company information from the chat. Based on the basic informations, it then performs Google searches to gather detailed company data required for the complex business rules to classify leads, calculate lead scores, and categorize potential leads. The chatbot also writes personalized messages based on recent milestones of the user's company found online for better engagement and tailored outreach. Then, it showcases similar use cases relevant to the user's expertise request and industry that the consultancy has successfully delivered, providing evidence of past successes and relevance to the user's needs. >> Stack: CrewAI, OpenAI, Python, FastAPI - Summary Generation and Insight Extractor of Financial Statements in Excel : Developed and implemented a Generative AI solution to automate the summarization of complex financial statements in Excel for one of Germany’s top banks. The solution provided in-depth insights, including detailed analyses of subtables, concise summaries of individual components, and a comprehensive overview of the entire document. This innovation transformed complex financial data into clear, actionable insights, significantly enhancing data comprehension and supporting better decision-making. The solution enabled faster processing of intricate datasets, empowering users to focus on strategic analysis rather than manual data aggregation. >> Stack: Langchain, OpenAI, Python

Enterprise Search: Contributed to a major global market research leader in the U.S. by developing an Enterprise Search application enhanced with Generative AI. As part of a big data and development team, I'm responsible for the integration of generative AI within the application. Responsibilities: -- Generate and curate synthetic datasets with generative AI to fine-tune Large Language Models (LLM). -- Fine-tune Gemini flash and open-source SLM (Small Language Models) for precise multi classification tailored to the enterprise search needs. -- Develop the RAG component of the application, capable of handling complex structured files at large scale.

Remote to US & LATAM >> Role : ---- Responsible for providing expert consultation in Generative AI initiatives for multiple clients of the company and develop projects to showcase to clients how to use Generative AI effectively in their business. >> Projects: --Proxxima AI for Marketing Strategy > Hyper-Personalization based on Product Review V1 : Used ChatGPT API + Langchain to develop an Application that : > Generates a summary of the reviews of selected products > Generates a Persona using selected reviews of a customer > Creates personalized engagement messages based on customer-selected reviews V2 : Migrated application to integrate with Vertex AI (PaLM2 - Bison ) + Langchain > Created and curated the training dataset of English product reviews based on a sample of the original Portuguese dataset. > Translated all the application to English and refactored scripts for the new data integration > Modularized Model integration for easier flexibility, allowing for changes in the future if necessary. Added Error handling and logging to the scripts. > Generate summary, personas and hyper-personalization engagement emails. >> Other Solutions: Developed the solution for the Pinecone Vector Database migration to the client's organization, by upserting over 500k vectors of dimension 1536 by interacting with the company's and the client's Pinecone APIs. | Stack : Langchain , LLM , ChatGPT API , Vertex AI , GCP , Bison-PaLM2, LLM Function Calling , Multi-tool Agents, Customized tools , Multimodal applications , Prompt Engineering , Poetry , Python , Streamlit , Docker , RAG - Vector Database, APIs , Websites.

• Architected scalable solutions using AWS : Designed end-to-end data pipelines and data intensive systems to efficiently extract, transform, and load large-scale datasets. Prepared technical and non-technical presentations to effectively communicate project architecture and benefits to final clients. >> S3 , Glue , Lambda , SQS , SNS , RDS - Postgres , Redshift , Python , Spark , SQL , Great Expectations , Data lake , Data warehouse • Orchestrated and guided the team in creating effective strategies to build a reliable and scalable system, using data engineering and software engineering best practices: Automated tests (unit , integration tests) as part of the CI/CD pipeline, IaC, Data modeling to optimize system overall performance, Version control , Techical documentation (architecture diagrams, data models, system dependencies), Error handling , Logging. • Implemented Data Governance practices: Designed and implemented data governance frameworks, including data lineage tracking, data quality validation, and data privacy controls. Included testing strategies within the data governance framework to verify data quality and integrity • Designed code patterns for maintainability and scalability: Developed robust code design patterns aligned with industry best practices, emphasizing code quality, reusability, and scalability. Implemented data quality validation techniques to ensure the accuracy and reliability of data. • Be the Technical leader of a team of senior data engineers, fostering a collaborative environment and empowering them to deliver high-quality work. Coordinated data governance initiatives, including data lineage, data quality validation and data modeling. • Collaborated with stakeholders and cross-functional teams: Worked closely with stakeholders, including business analysts and solution architects, to understand requirements and translate them into scalable data solutions , to ensure seamless integration and alignment across the project.

• Propose and deliver End-to-End Data Solutions for CRM team in Brazil and BI team in the US - Machine Learning, dashboards, reports, automation pipelines, analytics - based on business needs and requirements from stakeholders. • Lead teams of data scientists, data engineers, and data analysts in designing and implementing Data Solutions, from research and architecture to development and deployment, serving as the technical reference for the team - Data Solutions Architecture, Agile SCRUM . • Develop and maintain data pipelines, from data ingestion in all layers of the data lake to serving it to Data Solutions - AWS Glue, Databricks, Tableau, PowerBI, Snowflake, Docker, AWS, Python, Pyspark, SQL, APIs. • Drive changes to improve the efficiency and organization of the entire data team - Data Governance and Data Lineage processes. Main Achievements: - Proposed and led a team of 5 data scientists in the implementation of Customer Segmentation Analytics using RFM, LTV, and Product Analytics for all US customers. The results were used by the Marketing and E-commerce teams to build the strategy for the year 2023, then adopted by the brazilian CRM team - Troubleshooting of data pipelines related to data modeling and performance - Time zone issues, data modeling performance optimization ; - Proposed a new data modeling approach for strategic dimensions, focusing on maintaining historical data change to enable more accurate trend analysis - SCD type 2 - Proposed and implemented a new data modeling approach for dashboard data, addressing a key stakeholder pain point related to dashboard performance. - Implemented a manual Data Lineage for CRM resulting in a significant decrease in onboarding time for new data team members, from 2 weeks to 2 hours. - Proposed data governance and lineage strategies, advocating for increased visibility and insights for the data team - Automated solution to replace manual process. Responsible for the PoC of data lineage products .

Somos is a legal tech Startup that aims to fight for the dignity of people around the globe that suffer from injustices, helping law firms manage and analyze their Data of collective claims. I was responsible for creating a solution so that our deliverable would be more elegant, scalable and highly available than an Excel Sheet. Using the Business Acumen from the past months of work, I suggested a solution that could allow our lawyer clients to simulate negotiation scenarios with the counterpart, get the Data Analysis of those scenarios in real-time, filter the data to focus on specific features and export/share their results with their teams. My first key role was to perform the POC of the core tools to build this product. After a successful presentation to the CTO , my second key role was to put together and train a cross-functional team to build and productize the full solution. • Stack: Streamlit, GIT (Bitbucket), Docker, Docker Compose, Virtual Environments, Plotly, Python, AWS ECR , ECS Fargate , RDS , Lambda , SQS , S3 etc. My third key role is to research the market best practices on Data Products Development and Architecture, in order to capacitate our team to build competitive products. • Trainings : | Virtual Environment : What and Why | Code Version Control : Git | Containers : Docker | Containers Orchestration : Docker Compose

I started at Somos as a Senior Data Scientist , cleaning and organizing data ( ETL ) , creating insightful data analysis with Python and being responsible for the Data Strategy of the Questionnaires of the collective claims, working side by side with our international clients at PGMBM in London - UK . - Developed and implemented a data strategy using NLP , FuzzyWuzzy logic and analytics to identify and analyze clusters of victims of Brumadinho's Dam Collapse in order to compute how much each victim is entitled to according to business rules. - Developed and led the implementation by the Data Engineering team through the Data Strategy to analyze the big data scenario at the Uber Litigation, dealing with Uber's data of all events of each trip of 1,200 drivers for the last 5 years. • Stack : AWS Sagemaker, AWS Athena , AWS S3, Python , SQL , Machine Learning , Natural Language Processing (NLP), Pandas, Numpy, Scikit-Learn, Matplotlib, Seaborn , Jupyter Notebook , git (Bitbucket) Also, I'm responsible for being the technical reference by my teammates regarding Data Analysis and Data Science, as well as Data Strategy decisions. My third key role is to elaborate and create internal trainings to my team to explain our new implemented working method. • Trainings : | Agile Methodologies - SCRUM vs. KANBAN | KPIs, OKRs and Sprints I also conduct technical interviews of Senior Data Scientists and Data Analysts candidates, in both Portuguese and English .

iFood is a Brazilian on-demand food delivery platform Startup that became a unicorn in November 2018. My main responsibility is to act as the bridge between my business analyst team and iFood's Data team, as I'm the only one with a Data Scientist background. 》Data Analysis and ETL 》Machine Learning solutions 》Process Mining 》Identify Business Improvement opportunities with Data Science 》Stack : Python, PySpark, Databricks, SQL, Tableau, DriverlessAI (H2O autoML), Celonis EMS. 》Other Roles : • Guild Data with Stats : Being part of the organization, I was responsible for the development of the data driven culture by showing the fundamentals of statistics and machine learning to all collaborators, organizing the meetings, topics, recommending materials and articles. • Founder of the Debates group: Bi-weekly meetings to discuss articles about Business, Growth, Data driven culture and Professional development in the current market, promoting integration among areas and seeking innovative solutions for the Business. Main Sources: Towards AI, Medium, MIT Technology Review, Harvard Business Review etc.

》Data Analysis with Python 》Machine Learning models 》ETL with SQL

Data Wrangling (cleansing, enrichment, transformation), Data Analysis, Data Visualization , BI Reports. | Stack: Python, SQL, pandas, matplotlib, seaborn, jupyter notebook

- Main Achievements: • Machine Learning - Stanford University • Machine Learning Professional Certificate - IBM • Applied AI Professional Certificate - IBM • Data Science Professional Certificate - IBM • Artificial Intelligence Formation - Data Science Academy • Industry 4.0 Engineering Specialization - UNOPAR • Python for Data Analysis - Data Science Academy • Artificial Intelligence Fundamentals - Data Science Academy • Python Essencials - Python Institute + Cisco || Professional Badges || https://www.youracclaim.com/users/bruna-lemberck || GitHub || https://github.com/lemberck At the beginning of 2020 I decided to focus on Innovation. In order to research what is Innovation nowadays, I took a Specialization in Industry 4.0 Engineering, where I found out a great passion for Big Data and Cognitive Computing. From that time until present days, I've been studying and practicing on projects to work on my Machine Learning, Artificial Intelligence and Data Science skills, and decided to start a career in these areas in 2021. I've worked on over 30 projects, and some of them can be found here in my Linkedin and also here : https://github.com/lemberck

|| Cloud Computing Foundations || Cloud Virtualization, Containers and APIs || Cloud Data Engineering || Cloud Machine Learning Engineering and MLOps

This 11 weeks course provides a broad introduction to machine learning, datamining, and statistical pattern recognition, including the Best practices in machine learning (bias/variance theory; innovation process in machine learning and AI). The course will also draw from numerous case studies and applications, so that you'll also learn how to apply learning algorithms to building smart robots (perception, control), text understanding (web search, anti-spam), computer vision, medical informatics, audio, database mining, and other areas. || Supervised Learning - Linear Regression, Logistic Regression, Neural Networks, SVMs || Unsupervised Learning - Kmeans, PCA, Anomaly Detection || Special Applications - Recommender Systems, Large scale Machine Learning || Building a Machine Learning - Bias/Variance , Regularization, Evaluation of Algorithms, Learning Curves, Error Analysis, Ceiling Analysis.

| 400h | • 3D Metrology – Digitization • Smart Sensing Technology – Intelligent Manufacture Systems • Traceability : IoT applied to Production Management • Interactions between Big Data and Cloud Computing • Cyber-Physical Systems • Collaborative Robotics • PM Mind Map • Lean Principles • Product Design : MVP - Minimum Product Viable , Business Model Screen , Personas • Scientific Research Methodology

Expected: DEC/2020 To understand the newest manufacturing technologies, this specialization will provide a foundation in how digital advances are changing the landscape and capabilities of factories. Nine courses – developed with input from the manufacturing industry – touch on Industry 4.0 and its components, including digital manufacturing and design practices, the concept of the digital thread, the Internet of Things and Big Data. 1. Digital Manufacturing & Design 2. Digital Thread: Components 3. Digital Thread: Implementation 4. Advanced Manufacturing Process Analysis 5. Intelligent Machining 6. Advanced Manufacturing Enterprise 7. Cyber Security in Manufacturing 8. MBSE: Model-Based Systems Engineering 9. Roadmap to Success in Digital Manufacturing & Design and Industry 4.0 - Final Project

| 360h | • Industrial Instrumentation • Offshore Technology • O&G Support Systems • O&G Processing and Production Systems • Subsea Engineering | Reservoir Engineering | Prospecting | Well Engineering • Safety in Offshore Oil Production Units • Structural Behavior of Offshore Systems • Oceanic Environment • Naval architecture

| 4266h | 5 years | Bachelor's Degree in Industrial Automation and Control Engineering