Staff ML/AI Engineer · DataArt
Overseeing architectural strategy, technical standards and cross-team execution for AI/ML systems serving major corporate financial institutions.
Staff ML/AI Engineer @ DataArt
Ruben Esteche
I design and ship production AI systems — from RAG platforms and conversational agents to the cloud infrastructure that keeps them alive. Physicist by training, engineer by craft.
$ ▊
01
From wave functions
to production AI
Hello! I'm Ruben — a Staff ML/AI Engineer at DataArt, where I steer architectural strategy and cross-team execution for AI systems serving major financial institutions.
My path started in theoretical quantum physics — researching arrival-time and tunneling-time distributions at the Federal University of Pernambuco. The same instinct that drew me to modeling nature's uncertainty now drives me to build systems that reason over language: RAG platforms, conversational agents, and metadata-extraction pipelines running in production for companies like Bosch, NTT Data and Dell.
I care about the full lifecycle — not just the model, but the ingestion layers, vector & graph databases, CI/CD pipelines and cloud infrastructure that make AI reliable at scale. When I'm not shipping, I write about ML and physics on my blog.
0+years in data & ML
0companies, startup → enterprise
0languages spoken
02
Experience
Seven years shipping data and AI products — from junior data scientist to staff-level architecture ownership.
Senior ML/AI Engineer · DataArt
Built an AI agent that turns institutional databases into conversational applications — chain-of-thought structured responses, source tracking and interactive financial plotting. Designed metadata-extraction pipelines and hierarchical chunking systems natively integrated with AWS Lambda ingestion, over vector and graph databases.
Senior ML/MLOps Engineer · Bosch AI Hub
Architected production conversational AI, enterprise chatbots and RAG platforms with LLMs and deep NLP. Engineered scalable Azure environments with CI/CD, Docker and Kubernetes — cutting development lifecycle time by ~50%. Collaborated on computer-vision product design.
Lead Data Scientist & MLOps Engineer · NTT Data
Led the project's AI division — sprints, backlog, team performance and algorithmic bottlenecks. Shipped an enterprise NLP chatbot system reaching 92% classification precision, built generative AI-as-a-Service on microservices, and governed end-to-end MLOps architecture compliance.
ML Researcher & Backend Developer · Dell Lead
Web scraping, text classification and custom ML for a global-scale web accessibility initiative. Integrated NLP modules into high-throughput RESTful microservices; improved training/inference speed by 10% and compute-efficiency by 15%+.
Data Scientist · Sugestão Consultoria
From junior to technical lead of data-intelligence initiatives: predictive architectures with a 90% accuracy floor driving ~10% bottom-line expansion for clients, behavioral demand-forecast models at 95% accuracy, and automated ingestion pipelines.
03
Selected projects
Open-source work spanning generative AI, NLP tooling and computational physics.
◈
★ 11
Stable Diffusion API
Full backend for generative image creation — a REST API wrapping Stable Diffusion for unique image synthesis, ready for production integration.
◇
★ 2
Google Scholar Bot
Telegram bot that manages connections to Google Scholar and lets you interact with academic search results conversationally.
∿
★ 3
Physical Simulations
A series of physics simulations in C and Python exploring numerical methods for differential equations — from wave packets to chaotic systems.
⬡
★ 0
BERT-nha
A study case of BERT and LLMs in general through Noronha — dissecting transformer internals with an MLOps-first workflow.
◎
★ 1
LinkedIn Analysis
Data analysis of my own LinkedIn network — graph exploration and visualization of professional connection patterns.
✦
★ 1
Manim Studies
Educational math animations built with Manim — turning abstract mathematics into beautiful, shareable visual explanations.
04
Stack & domains
The tools I reach for when taking AI from notebook to production.
🧠
AI · NLP · LLMs
Building systems that understand and generate language, grounded and observable.
- RAG
- LLM Agents
- Transformers
- Generative AI
- Deep Learning
- Computer Vision
- CNNs
- Statistical Modeling
⚙️
MLOps & Infra
Reliable pipelines from ingestion to inference — automated, versioned, reproducible.
- Docker
- Kubernetes
- CI/CD
- IaC
- Azure DevOps
- Vertex AI
- DataOps
- Vector & Graph DBs
☁️
Cloud Platforms
Multi-cloud production experience across the three major providers.
- AWS Lambda
- ECS
- S3
- API Gateway
- GCP
- Microsoft Azure
🛠️
Backend & Languages
Solid engineering foundations under every model I ship.
- Python
- FastAPI
- Flask
- Django
- SQL
- C / C++
- Redis
- MongoDB
- ElasticSearch
Certifications
-
GCP
Professional Machine Learning EngineerGoogle Cloud · 2025
-
GCP
Build & Deploy ML Solutions on Vertex AIGoogle Cloud · 2025
-
GCP
Generative AI FundamentalsGoogle Cloud Skills Boost · 2023
-
ICTP
School of Research Data ScienceICTP-SAIFR · 2022
-
IBM
Databases & SQL for Data ScienceIBM · 2022
-
UFC
CNN Design with TensorFlowFederal University of Ceará · 2020
05
Research & writing
Before LLMs, I modeled time itself — quantum foundations research that still shapes how I think about uncertainty.
A Space-Time Symmetric Extension of Quantum Mechanics
Developed a new space-conditional solution for particles under time-dependent initial conditions, benchmarked against absorptive-potential models and probability-current time distributions. Presented at national physics conferences and defended at UFPE.
Fluid Interfaces: Curvature Relations & Euler Elastics
Revisited Euler's elastica to analytically describe curvature relations in two-fluid immiscible systems, complemented by numerical simulation of nonlinear ODEs and laboratory experiments.
REsteche's Lab Notebook
Educational posts where physics meets machine learning — deep-learning text projects, gaussian wave packets, Lorenz attractors, statistical mechanics and more, all as executable notebooks.
06
Education
M.Sc. Theoretical Quantum Physics
Federal University of Pernambuco
Minor in Computational Physics. Thesis on arrival & tunneling time distributions in a space-time symmetric formalism. CAPES graduate fellowship.
Postgraduate Degree, Data Science
Descomplica University
Specialization covering statistical learning, ML engineering practice and applied data science.
B.Sc. Physics
Federal University of Pernambuco
Undergraduate research in fluid dynamics and Euler elastics; teaching assistant for Physics for Computer Science.
07
Let's build something
intelligent together
Collaborations, consulting, speaking, hiring — or just to talk shop about LLMs and quantum mechanics. My inbox is open.