Skip to content
View aaitorm's full-sized avatar
  • Student at UPV/EHU
  • Spain

Block or report aaitorm

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don't include any personal information such as legal names or email addresses. Markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
aaitorm/README.md

Hey, I'm Aitor

AI Engineer · Machine Learning · LLMs · Computer Vision

LinkedIn · Email · GitHub


About me

I'm an Artificial Intelligence student at the University of the Basque Country (UPV/EHU) focused on building and evaluating machine learning systems in real-world scenarios.

My work centers around:

  • Machine Learning Systems — end-to-end pipelines, training and evaluation
  • LLM Evaluation — truthfulness, reliability, prompting strategies
  • Computer Vision — object detection, model comparison

I'm particularly interested in model behavior, evaluation, and reliability under real-world conditions.

I'm currently looking for AI Engineering internships or collaborations.


Tech Stack

Languages

Python

ML / DL Frameworks

PyTorch Transformers scikit-learn

Data & Notebooks

NumPy Jupyter

Tools

Git FastAPI LangChain


Projects

Optimization of graph partitioning using GRASP, Simulated Annealing and Genetic Algorithms.

  • Compared solution quality and computational cost
  • Analyzed convergence behavior across algorithms
  • Implemented all methods from scratch

Optimization Heuristics Python


Fine-tuning diffusion models for personalized image generation.

  • Prompt-based identity control using custom token
  • Built FastAPI backend for generation pipeline
  • Integrated external APIs (Replicate)

Diffusion Models Computer Vision FastAPI


Evaluation pipeline to measure LLM truthfulness and informativeness.

  • Built few-shot + LLM-as-a-judge framework
  • Multilingual evaluation (Spanish / Basque)
  • Compared base vs instruction models

LLMs Evaluation Prompting


Object detection pipeline comparing CNN and Transformer-based models.

  • Trained and evaluated on MPII dataset
  • Compared RT-DETR vs CNN baselines
  • Performed qualitative and quantitative analysis

Computer Vision Object Detection Transformers


Agent-based system integrating tools and reasoning workflows.

  • Multi-step reasoning with tool usage
  • Integration with external knowledge sources
  • Response synthesis pipeline

LLMs Agents LangChain


Current Focus

  • LLM evaluation and reliability
  • Applied computer vision systems
  • Optimization methods

Let's connect

I'm open to collaborations, internships, or interesting AI projects.

email: aitormilikua@gmail.com


Pinned Loading

  1. ai-photo-personalization ai-photo-personalization Public

    Personalized image generation using fine-tuned diffusion models with FastAPI backend and rate-limited API

    Jupyter Notebook 1

  2. truthfulqa-llm-evaluation truthfulqa-llm-evaluation Public

    Evaluation of Llama-3.1-8B Base vs Instruct on TruthfulQA using few-shot prompting and automatic judge models

    Python

  3. head-detection-mpii-rtdetr head-detection-mpii-rtdetr Public

    Head detection on MPII dataset using CNNs and Transformer-based models (RT-DETR, SimpleRTDETR)

    Python 1

  4. community-detection-heuristics community-detection-heuristics Public

    Heuristic and metaheuristic approaches (GRASP, Simulated Annealing, Genetic Algorithms) for community detection in complex networks

    Jupyter Notebook