Portrait of Andres Ramirez-Jaime

Hi there, I'm

Andres Ramirez-Jaime

Ph.D. Candidate, Electrical & Computer Engineering (University of Delaware - expected Jul 2026)

Generative/reconstructive ML for computational imaging & perception (diffusion models, transformers) across 3D LiDAR, hyperspectral remote sensing, and gigapixel pathology.

Ph.D. candidate focused on generative and multimodal ML for computational imaging and perception. I build diffusion/transformer pipelines for 3D LiDAR and hyperspectral sensing (NASA CASALS) and gigapixel pathology. Industry research engineering experience at Apple (LLM tool-use: RAG, LoRA, evaluation suites) and Vertex Pharmaceuticals (semantic segmentation for whole-slide imaging); I design reproducible data -> training -> evaluation workflows for large-scale experiments.

Download CV

01. Experience

Apple Inc.

Jun 2024 -- Aug 2024
Large Language Models and Generative AI Engineering Intern
  • Built a retrieval-augmented generation system grounded in the iPhone user manual, improving internal tool descriptions for ~85% of tools.
  • Fine-tuned Apple Intelligence LLMs for iPhone-focused tool usage via dataset curation and LoRA adapters; raised single-turn tool-selection accuracy from 61% to 90.2%.
  • Created evaluation datasets and hand-crafted test suites (~2,500 conversations across ~80 tools) to stress-test edge cases and diagnose failures.
  • Optimized the training/evaluation pipeline to increase iteration rate from 1 to >4 cycles/day, accelerating experimentation.
  • Reduced tool hallucinations (nonexistent tools) from ~3% to 0% on internal testing via error analysis and mitigation recommendations.

Vertex Pharmaceuticals

Jun 2023 -- Aug 2023
Computer Vision and Machine Learning Engineer
  • Partnered with biologists, physicians, and chemists on early drug discovery (IPF, ADPKD), translating scientific goals into ML deliverables and evaluation criteria.
  • Built a U-Net semantic segmentation model for gigapixel pathology images (>90% accuracy); replaced outsourced processing (100 slides; up to 3-month turnaround) with in-house inference (~15 min per processed region), avoiding up to $600k/year in external spend.
  • Developed an automated pipeline to segment kidney organoids in whole-slide images and estimate morphology (<3% error), improving measurement consistency for downstream analysis.

University of Delaware

Feb 2022 -- Present
Research and Teaching Assistant (NASA CASALS)
  • Lead research on generative and multimodal ML for LiDAR and hyperspectral remote sensing (diffusion models, transformers, GANs) within the NASA CASALS project.
  • Developed reconstruction and super-resolution methods for the HyperHeight Data Cube (HHDC) 2 m NEON dataset (~100k 3D tensors); achieved <1 m MAE on CHM/DTM under ~25% sampling.
  • Built end-to-end pipelines (data processing -> training -> evaluation) on ~60 GB datasets; ran ~5-day training cycles on a single A100 and produced inference over ~600 m^2 areas in ~12 min.
  • Published and presented research: 17 peer-reviewed papers (7 first-author) at international venues in remote sensing and computational imaging.
  • Mentored 2 Ph.D. students, 1 master's student, and visiting researchers; TA for Statistical Learning, Imaging and Deep Learning, and Probability/Statistics.

University of Delaware

Jul 2021 -- Dec 2021
Visiting Scholar
  • Designed the HyperHeight Data Cube (HHDC) representation for efficient storage and processing of compressed 3D satellite LiDAR data, enabling downstream learning and reconstruction workflows.
  • Implemented a 3D convolutional autoencoder for HHDC reconstruction, improving reconstruction quality vs. classical baselines by +6 dB on CHM and +18 dB on DTM.

02. Technical Skills

Programming

Python C MATLAB LabVIEW

ML & Deep Learning

PyTorch TensorFlow Keras Diffusion Models Transformers GANs LLM fine-tuning (LoRA) Retrieval-Augmented Generation (RAG) Multimodal Networks

Computer Vision / Imaging

Semantic Segmentation Super-Resolution Denoising Reconstruction Computational Imaging Medical & Remote-Sensing Imagery

Data / Systems

  • Large-scale training and evaluation (gigapixel images; 3D tensors; GPU training).
  • Dataset curation plus experiment evaluation and test-suite design.
  • Scalable pipelines spanning data processing -> training -> inference.

Tools

Linux Git GIS LaTeX

03. Datasets & Open Source

HHDC - Hyperheight Data Cube Denoising & Super-Resolution

Public dataset of ~89k 3D photon-count LiDAR cubes built from NEON discrete-return LiDAR for denoising and spatial super-resolution of forest canopies.

Hugging Face: anfera236/HHDC License: CC BY 4.0

04. Publications

Super-Resolved 3D Satellite LiDAR Imaging of Earth Via Generative Diffusion Models

Ramirez-Jaime, A., Porras-Diaz, N., Arce, G. R., Stephen, M.
IEEE Transactions on Geoscience and Remote Sensing, 2025

SpectralCam: High-Resolution Low-Cost Spectral Imaging Using DSLR Cameras

Paruchuri, A., Ramirez-Jaime, A. et al.
Proc. IEEE ICASSP, 2025

Super-Resolution of Satellite Lidars for Forest Studies Via Generative Adversarial Networks

Ramirez-Jaime, A., Porras-Diaz, N., Arce, G. R. et al.
Proc. IGARSS 2024

Transformer End-to-End Optimization of Compressive LiDARs Using Imaging Spectroscopy Side Information

Porras-Diaz, N., Ramirez-Jaime, A., Arce, G. R. et al.
IEEE Transactions on Geoscience and Remote Sensing, 2024

HyperHeight LiDAR Compressive Sampling and Machine Learning Reconstruction of Forested Landscapes

Ramirez-Jaime, A., Pena-Pena, K., Arce, G. R. et al.
IEEE Transactions on Geoscience and Remote Sensing, 2024

The Development and Implementation of a Low-Cost Mechanical Ventilator in a Low-Middle-Income Country During the COVID-19 Pandemic: The Unisabana-HERONS

Giraldo-Cadavid, L. F., Echeverry, J., Varon, F., Ramirez-Jaime, A. et al.
Heliyon, 2024

Full publication list on Google Scholar →

05. Education & Leadership

Education

Ph.D. in Electrical & Computer Engineering

University of Delaware

Expected Jul 2026 | GPA: 3.92

2022 George W. Laird Fellow; 2024 ECE Signal Processing Award; 2024 Doctoral Fellowship for Excellence.

M.S. in Computer & Electronic Engineering

University of Los Andes

Mar 2016 | Graduated Cum Laude

B.S. in Electronic Engineering

University of Los Andes

Oct 2013

Teaching & Leadership

Jan 2018 -- Dec 2021

Mechanical Engineering Professor

University of La Sabana - Chia, Colombia

  • Supervised 12 undergraduate thesis projects; taught courses in prototypes and manufacturing.
  • Designed and programmed embedded control for the UNISABANA HERONS mechanical ventilator deployed in 400+ units during COVID-19; implemented respiratory control and data collection in C and LabVIEW to support regulatory approval.
  • Led a robotics team for RoboCup 2019 (Sydney); reached the Community Shield final as runner-up, owning perception, control, and systems integration.