Hi! I'm Gabrielly, but you can call me Kimi 🧩
I'm an ML Engineer focused on MLOps and AI Infrastructure with 2 years of experience. I build production AI systems that scale and cost less.

I'm pursuing a Bachelor of Science in Science and Technology with focus on Applied Computing at Federal University of Rio Grande do Norte. Previously completed 1.5 years of an Associate of Science in Data Management at Federal University of Piauí (GPA: 9.5/10), where I published two papers.

I'm currently building Billink AI — a SaaS platform that automates cloud and AI infrastructure cost auditing using multi-agent LLM pipelines.

What it does:

• Analyzes billing data from AWS, OpenAI, and Anthropic
• Identifies waste patterns (idle GPUs, expensive model overuse, storage inefficiencies)
• Generates actionable recommendations with projected savings

Roadmap:

• ✅ Architecture designed (3-stage agent pipeline: Extractor → Analyst → Advisor)
• 🚧 MVP in development (CSV upload → analysis → PDF report)
• 🎯 Goal: 2 paying customers for validation

Follow the journey: I share progress, learnings, and challenges on LinkedIn.

👩‍💻 Currently, I am...

• 💼 ML Engineer / MLOps Engineer at Cortex: Architected an Agentic AI system processing 80K+ stores across 700+ malls on Kubernetes, enabling a $2M revenue product line. Implemented monitoring, alerting, and batch inference pipelines.
• 🤖 AI Researcher at UFPI: Building hybrid BiLSTM-Transformer models for hate speech detection with feedback loop for retraining. Fine-Tune Baseline | BiLSTM Baseline

🛠️ Tech Stack

• MLOps & Infrastructure: Kubernetes (EKS), Karpenter, Spot Instances, Docker, MLflow, CI/CD for ML, Model Monitoring, Drift Detection
• Cloud: AWS (S3, Lambda, EC2, EKS, DynamoDB, Cost Explorer), Terraform, FinOps
• Machine Learning: LLMs, Agentic AI, RAG, NLP, TensorFlow, PyTorch, Physics-Informed ML
• Data Engineering: PostgreSQL, ClickHouse, Supabase, ETL/Data Pipelines, Analytics Engineering
• Observability: Grafana, CloudWatch, HyperDX (P95 Latency, Error Budgeting)
• Programming: Python, SQL, PySpark

✨ Highlights

• 🏆 Transpetro/Petrobras Hackathon (7th place): Physics-Informed ML for industrial prediction — identified $1.89M in fuel savings and 1,198 tons CO2 avoided. Paper | Dashboard | Model | REST API
• 🇧🇷 I Speak Kanoê: Built the first dataset for Kanoê indigenous language preservation. Documentation | Dataset
• 📊 Social Impact: Created Water Inequality Dataset of NE Brazil and published peer-reviewed research. Paper | GitHub
• 📊 Cloud Infrastructure Analyst (Freelance): Audited infrastructure handling 3.26B requests with 99.92% availability during Black Friday, analyzing EKS cluster efficiency and Spot Instance resilience.

🎖️ Certifications

• Stanford & DeepLearning.ai: Machine Learning Specialization. ML Project
• Harvard Aspire Institute: AI-Integrated Leadership Program (AILP) Alumna
• DataCamp: AI Engineer Associate & Data Engineer Associate

📚 My "Stanford Roadmap"

• As a neurodivergent learner, I thrive on structure and depth. I am executing a rigorous, self-directed curriculum based on Stanford University's Computer Science BS parallel to my formal degree.

🌍 Leadership & Volunteering

• 🧑‍🏫 Teaching Fellow: Data and AI for Social Analysis course at UFPI
• 🧑‍🏫 AI Instructor: Teaching AI for women entering tech at NGO
• 🇺🇳 United Nations MGCY: Major Group for Children and Youth volunteer

💬 Let's connect

• 💼 LinkedIn: Gabrielly Gomes
• 📧 Email: gabrielly.gomes@billinkai.com