ML Engineer · Data Systems · GenAI

ML systems that
survive production.

End-to-end ML from data to deployment: credit, fraud and AML, time series, and agent-based systems with harnesses, RAG and fine-tuning. Calibration, reproducible evaluation and leakage control as constraints, not afterthoughts. Six public systems, real numbers.

Python · Scikit-learn · LightGBM · XGBoost · Polars Docker · Linux · uv · FastAPI · MLflow Fraud · Credit Risk · AML · Time Series RAG · AI Harness · Fine-tuning · LoRA · MCP · AI Agents ASIR background
github.com/cacelass linkedin.com/in/cacelas
$
about Engineering identity
Approach

I treat ML problems as systems problems first. A model that can't be reproduced or re-deployed is a liability. My workflow enforces clean separation between ingestion, feature computation, training, and evaluation, by structure rather than convention. In finance that discipline is what makes a system auditable.

Systems background

Trained and worked as a Systems Technician (ASIR). What drew me most to the field was databases: schema design, query writing, index tuning and execution plan analysis. That background shapes how I think about every ML system: infrastructure, environment isolation, and failure modes before model architecture.

Production awareness

I design pipelines that don't break when schemas change, models that return calibrated probabilities rather than raw scores, and evaluation setups that catch leakage before it reaches production. I know the difference between a metric that looks good and a decision that holds up out-of-sample.

Passion

I'm passionate about data inside and outside work. I love learning, taking on challenges and getting better at what I do. My professional aspiration is to build decision systems that real institutions trust, in credit, fraud and investment, while keeping every project better than the last.

$
projects Selected systems
more Further work
Two-level intrusion detection (LightGBM + KMeans) on CIC-IDS2017 with SMOTE and MLflow tracking.
security
Churn probability for customer data with a modular RF pipeline and feature engineering.
churn · ML
Unsupervised clustering of 167 countries by socioeconomic indicators.
clustering
Online frequency-based predictive agent for Rock-Paper-Scissors.
agents
In development: a portfolio manager system to decide when to reinvest and when to withdraw, built on the evaluation rigour of Stock-Market-Prediction.
$
stack Capability model
Core engineering
Python SQL PL/SQL Hive Linux Docker Git Bash uv FastAPI Nginx
ML / Data pipeline
Scikit-learn LightGBM XGBoost PyTorch Polars Pandas NumPy MLflow Optuna Matplotlib Seaborn
Cloud / Infrastructure
Azure MySQL SQLAlchemy Apache NoSQL Power BI
Azure Data Fundamentals Power BI / DAX ASIR
Domain knowledge
Fraud detection Credit risk Churn modelling Time series Clustering Calibration Leakage detection Conformal prediction
AI engineering
Generative AI RAG AI Harness LoRA Fine-tuning MCP AI Agents
What I can build
End-to-end supervised pipelines (classification + regression) Fraud and AML detection with imbalance-first evaluation Credit scoring with calibrated probabilities and decoupled thresholds Reproducible project environments with versioned dependencies via uv Evaluation frameworks with explicit data leakage controls Time-series models with temporal validation and out-of-sample backtest Containerised ML services (Docker + FastAPI) for cloud or on-premise Drift monitoring for models in production RAG systems with verifiable sources and guardrails AI harnesses that enforce quality gates in code Multi-agent systems with deterministic Python agents LLM fine-tuning and LoRA adaptation
$
reading Books that shaped how I work
Cathy O'Neil. Why opaque, unregulated models amplify inequality — and why model risk is a first-class concern in finance.
ethics · model risk
Agrawal, Gans & Goldfarb. The economics of prediction: what AI makes cheap and how it changes business decisions.
economics · AI
Chip Huyen. The reference on turning models into reliable systems: deployment, monitoring, drift and iteration.
ml engineering
Philip K. Dick. The classic on what it means to be intelligent, and why interfaces should feel like people.
fiction · AI
Ludovic Slimak. Reconstructing behaviour from partial evidence — the same discipline as working with incomplete data.
curiosity