Open to opportunities

Miguel Dias

@migueldias

Message

MLOps engineer building reliable, observable ML platforms and scalable LLM inference.

Germany

Message

What I'm looking for

I’m looking to build and run production ML platforms as a product—end-to-end pipelines, orchestration, and LLM serving with strong observability. I want teams that value reproducibility, cost control, and developer-friendly tooling.

I’m an MLOps engineer with 8+ years building production ML systems, focused on the pipelines, orchestration, and platform tooling that let ML teams ship reliably. I build end-to-end ML pipelines and developer-facing platforms—shared Python libraries, experiment tracking and model registry with MLflow, monitoring/logging/tracing with Langfuse, Grafana, and Prometheus, plus drift detection and reproducibility—on Kubernetes across GCP and AWS.

I treat the platform as a product for ML engineers, with a sharp eye on cost and repeatable deployments. In my recent work I modernized an ML platform (owning CI/CD and shared libraries), cutting code duplication and setup time by 95% and accelerating deployments by 80%, and I led LLM-based systems that reduced inference costs by ~99% and response latency from ~2 minutes to ~15 seconds. I also lead data/ML projects end-to-end, including LLM RAG extraction and data/algorithm pipelines deployed to Google Cloud.

Experience

Work history, roles, and key accomplishments

Current

Machine Learning Engineer

Current

Radancy

Aug 2025 - Present (11 months)

Modernized the ML platform by owning deployment infrastructure, CI/CD workflows, shared DS libraries, and migrating tooling to GitHub, delivering large reductions in duplication and setup time. Led end-to-end development and deployment of LLM-based applications, building scalable inference pipelines and observability, and optimizing cost and latency using KEDA/KServe with LiteLLM/vLLM and AWS Bedr

KEDA KServe MLFlow Langfuse CI CD GitHub LiteLLM VLLM Observability AutoScaling Monitoring Prompt Engineering

Machine Learning Engineer

Vodafone GmbH

Aug 2024 - Aug 2025 (1 year)

Redesigned the MLOps framework on Google Cloud, reducing codebase size and complexity while accelerating scheduling and feature deployment. Built an LLM-based Retrieval-Augmented Generation (RAG) project using customer transcripts and collaborated to improve data quality with topic modeling.

Google Cloud MLOps CI CD Retrieval Augmented Generation RAG Topic modeling LLM Data Quality

Senior Data Scientist

Vodafone GmbH

Mar 2022 - Jul 2024 (2 years 4 months)

Developed and led a QGIS plugin for fiber rollout planning to improve data access and efficiency for network planners. Created and maintained a universal multi-sector dataset and led integration of algorithms and pipelines to Google Cloud, optimizing code with parallelization to reduce compute time and costs.

QGIS Google Cloud Data Pipelines Parallel Computing Cost Optimization

Data Scientist

Vodafone GmbH

Apr 2020 - Feb 2022 (1 year 10 months)

Migrated network coverage calculations to Google Cloud, significantly reducing computation time and enabling dynamic analysis. Optimized fiber node recommendation using graph techniques, migrated geoprocessing to PostGIS on Google Cloud, and built clustering models to identify saturated cell towers.

Google Cloud PostGIS Clustering Models

Big Data & Data Science Trainee

Vodafone GmbH

Oct 2018 - Apr 2020 (1 year 6 months)

Built scalable ETL pipelines with Apache Spark and improved revenue models by integrating geospatial data. Led a crowdsourcing-based drivetest reconstruction project to reduce manual data gathering effort and CO2 emissions.

Apache Spark ETL Geospatial Data Data Integration Crowdsourcing

Machine Learning Research Intern

INESC-ID

Feb 2018 - Jul 2018 (5 months)

Researched and developed secure data mining schemes for speech signals (paralinguistics), using cryptographic primitives alongside neural networks and SVMs. Produced emotion predictions with 82% accuracy in a fully secure, private environment.

Cryptography Neural Networks SVM

Data Analytics & AI Intern

Energias de Portugal (EDP)

Feb 2017 - Jan 2018 (11 months)

Implemented a search-based monitoring platform to identify the origin of scalable failures and crashes using machine logs. Developed an ML anomaly detection algorithm to predict failures and crashes from machine logs, improving productivity and customer satisfaction.

Monitoring Anomaly Detection