THIAGO GUIMARAES
Senior Software Engineer
Real-Time Drilling Analytics
AI Platforms
ABOUT
Senior Software Engineer with 10+ years building cloud-native, real-time drilling analytics platforms and machine learning prediction systems for Shell, Aramco, and ADNOC.
Led ADNOC's RTOC platform implementation across 120+ rigs, designed a hybrid AI-Physical stuck pipe prediction system with 90%+ accuracy and 3 SPE/IPTC papers, and engineered rig state detection, drilling KPI dashboards, and real-time surveillance systems across onshore, offshore, and island operations.
Currently applying real-time data engineering and AI platform expertise to cloud-native SaaS — Python, React, AWS, event-driven microservices. M.Sc. in Data Science & AI.
Backend &
Data
Engineering
Building production-grade APIs, event-driven pipelines, and stream processing systems that handle high-frequency sensor data at scale — from WITSML ingestion to real-time analytics warehouses.
- Python
- JavaScript ES6+
- TypeScript
- Java 17+
- HTML5
- CSS3
- Node.js
- Spring Boot
- FastAPI
- Django
- REST APIs
- Microservices
- WebSocket
- Event-Driven Architecture
- Serverless
Frontend
Engineering
Interactive dashboards, real-time data visualization, and responsive interfaces for drilling operations and analytics platforms — from sensor time-series charts to multi-well fleet views.
- React.js
- Context API
- D3.js
- Canvas
- Data Visualization
- Responsive Design
Data &
Artificial
Intelligence
Feature engineering on sensor streams, ML model design for industrial prediction, LLM-powered analytics agents, and modern data stack orchestration — from raw wellbore data to actionable insights.
- SQL
- MongoDB (Time Series)
- PostgreSQL
- Stream Processing
- Feature Engineering
- scikit-learn
- TensorFlow
- Polars
- Pandas
- dbt
- Cube.js
- LangGraph
- LangChain
- OpenAI API
- OpenRouter
- Claude Code
- MLflow
Cloud &
DevOps
Infrastructure as code, container orchestration, and CI/CD pipelines powering real-time drilling platforms and cloud-native SaaS deployments across AWS.
- AWS Lambda
- S3
- SQS
- Kinesis
- AWS Batch
- Step Functions
- EventBridge
- ElastiCache
- SageMaker
- Docker
- Kubernetes
- Terraform
- GitLab CI
- GitHub Actions
Oil & Gas
Engineering
Deep operational knowledge in drilling analytics, real-time operations centers, and physics-informed prediction systems. Specialized in WITSML data, T&D models, and rig surveillance across onshore, offshore, and island operations.
- WITSML / ETP
- Real-Time Drilling Platforms
- Drilling Analytics
- Drilling KPIs
- Torque & Drag Models
- Hole Cleaning
- Rig State Detection
- RTOC Operations
- Predictive Analytics
Experience
Cloud-native data platform for fitness SaaS — real-time event pipelines, ETL orchestration, analytics warehouse, and AI-powered business intelligence.
- Built 19-module Python ETL system (SQLAlchemy, Pydantic) syncing data from external APIs into PostgreSQL and MongoDB, orchestrated via AWS Batch (Fargate Spot) with Step Functions dependency chains, EventBridge scheduling, and multi-tenant Terraform IaC deployment
- Replaced polling with real-time event-driven architecture: Spring Boot webhook gateway → SQS FIFO (per-entity ordering, dedup) → idempotent consumers with transactional outbox pattern, delivering data changes in seconds vs. minutes
- Reverse-engineered third-party webhook contract by profiling 5.96M events across 7 dimensions using Polars, producing an integration playbook that informed the real-time architecture design
- Built dbt + SQL analytics warehouse (staging → intermediate → marts) producing 50+ KPIs per month, served via Cube.js semantic layer with Redis caching for sub-second dashboard queries
- Built an AI operations agent (FastAPI, LangGraph) — 12 specialized tools delivering real-time coaching on trainer performance, client retention risk, and revenue optimization via streaming chat
- Developed fullstack admin dashboard (Django REST Framework, React) with staged rollout system using Redis allowlists, enabling production activation across 500+ clubs without code deployment
Real-time drilling data platform (Python, Java, React, MongoDB) serving Shell, Aramco, and ADNOC. Processed 5-second WITSML streams from 120+ rigs across onshore, offshore, and island operations.
ADNOC RTOC — Technical Implementation Leader (2023)
- Led end-to-end implementation of ADNOC's Real-Time Operations Center platform, serving as primary technical interface between Intelie and ADNOC Drilling + ADNOC Offshore (120+ rigs)
- Built real-time surveillance dashboards ingesting 5-second WITSML streams (hookload, WOB, SPP, RPM, torque, flow, pit volumes, gas) with interactive time-depth curves and multi-well fleet views
- Implemented Torque & Drag physical model monitoring with real-time hookload/torque event classification, friction factor anomaly detection with severity thresholds, and automated deviation alerts
- Developed rig state detection engine covering 17+ activity states auto-classified from surface sensor data, feeding connection time tracking, trip speed analysis, and NPT auto-detection
- Designed KPI comparison dashboards for well-vs-well, rig-vs-rig benchmarking with P10/P50/P90 offset envelopes and learning curve tracking
- Implemented KPI Data Quality SOP system with automated sensor validation, DDR-vs-real-time reconciliation, and per-channel quality scoring
Aramco EXPEC ARC — AI/ML Engineer (2021–2022)
- Designed and built a hybrid AI-Physical stuck pipe prediction system combining supervised ML with real-time outputs from T&D, Hole Cleaning, and Hydraulics physical models
- Engineered the real-time data pipeline: automated WITSML ingestion, data quality engine (gap detection, spike filtering, cross-channel validation), and intensive feature engineering at 5-second resolution
- Developed 3-class stuck pipe classifier (mechanical / differential sticking / normal) trained on 200+ wells, achieving 90.1% accuracy, 95.9% precision
- Built rig crew notification interface integrated with Aramco's RTOC, delivering real-time stuck pipe risk alerts with contextual parameter visualizations
- Co-authored 3 peer-reviewed papers (SPE/IADC Abu Dhabi, SPE Offshore Europe Aberdeen, IPTC Dhahran) — cited in University of Texas review (SPE-220725-PA, 2025)
- Built timeseries labeling platform (React + Python + MongoDB) for Aramco drilling SMEs across 500+ labeled events
Built real-time drilling analytics applications on a cloud-native streaming platform for Shell's global drilling operations.
- Architected real-time communication platform for Shell's drilling operations using server-side event-driven architecture, WebSocket streaming, and reactive React components, deployed across Shell's global rig fleet
- Built slide-drilling scoring microservice processing real-time directional drilling data to evaluate motor performance and directional efficiency