// Machine Learning Engineer · Data Scientist
Miguel Gutierrez
Machine Learning Engineer and Data Scientist with 4+ years across industry and research, deploying systems spanning reinforcement learning, time series, computer vision, generative AI, NLP, and LLM-powered applications. Currently completing an MSc in Mathematical Engineering (Statistical Learning) at Politecnico di Milano on a government scholarship (Top 5%). Background bridging economics, applied mathematics, and software engineering.
01 — About
02 — Selected Projects
Full-stack RAG + WebXR app for Milan's Piazza del Duomo: LLM assistant grounded in a Chroma vector database, AR treasure hunt, and vision-model photo verification. Containerized with Docker and deployed to AWS (EC2, ECR, S3, CloudFront).
Extended GAIR-RANSAC to decompose noisy point clouds into superquadric primitives, cutting average Chamfer distance by 40% (4.1→2.5×10⁻³) across synthetic scenes and 10 real Thingi10K meshes.
Accuracy-focused evaluation of open-source models for agentic data analysis tasks.
High-performance graph search with CUDA acceleration for OracleX × Polimi challenge.
Dual-stream text + image matching pipeline for e-commerce product deduplication.
Computer vision pipeline for 3D scene recovery from a single photograph.
03 — Writing
Interactive Blog STAG 2025
Fitting Geometry to Messy Point Clouds: Robust 3D Primitive Decomposition with Superquadrics
Drag the 3D figures to explore
Towards AI
Taking the Guesswork out of Anomaly Detection: Let Histograms Set the Threshold
Nov 26, 2023
LatinX in AI
Creating a Clever Code Interpreter Tool With Langchain Agents + Advanced Prompt Techniques
Oct 2, 2023
Medium
Quantile Mapping Bias Correction
Feb 22, 2023
04 — Research
A Model-Based Filter to Improve Local Differential Privacy
→ doi.org/10.52591/lxai202207171