David Villarreal Email & Phone Number

San Francisco, CA, US

Malta, NY, US

• Working on all product lines for the integrated circuits manufacturer, ranging from 130 nm nodes used for things like Internet of Things (IoT) and wearables, to 12 nm nodes developed by Intel and United.
• Responsible for manufacturing optimization through pattern matching to check for lithographic hotspots in an integrated circuit. Created Machine Learning workflows, using Neural Networks, along with own data extraction.
• Built a knowledge graph chatbots through creating ontology to be extracted by LLMs to be converted to Python lists, readable in Neo4j as a CSV. Embedded conversations with Neo4j Cypher Declarative Query Language and.

Malta, NY, US

• Created AutoML workflows, combining annealing and genetic algorithm workflows to create and select optimal neural network to avoid hyperparameter tuning. Working on 3DIC pin placement algorithm development off of.
• Set up and used a variety of AWS Services for the PDK and DFM Chatbots, including EC2, Lambda, Amazon Simple Queue Service (SQS), OpenSearch, Kendra, Bedrock, SageMaker, DynamoDB, and Amazon Elastic Container Registry.
• As the acting manager for the Job Enrichment Project for research projects development trained staff on AWS and created code for outlier detection, using multimodal capabilities for Virtual Assistants (VAs) and code.
• Performing extensive work with Deep Learning, developing neural networks for regression and classification problems. Created testing algorithms to interpret Neural Network learning, and Recurrent Neural Network (RNN).
• Created Python Master Code to simplify usage of available tools. Used all machine Learning packages, such as NumPy, SciPy, TensorFlow, Keras, PyTorch, Pandas, and Scikit-learn.

• Founding Engineer for the firm specializing in predicting valence of popular songs. Working as a Full Stack Engineer and mobile app developer.
• Developed data analytics algorithms. Using Django and GraphQL to create the statistical output to determine customer viability and increase ROI; Selenium for web scraping; PyTorch to create Neural Network regression.

Berlin, DE

• Utilized the code from Harvard professor Boriz Kozinsky’s GitHub repository (NequIP), meant for building E(3)-equivariant interatomic potentials. Created equivariant neural networks for SrTiO3 (Strontium Titnate).
• Created SrTiO3 crystal in LAMMPS (Molecular Dynamics Language) using object orientation and ran structure relaxation for molecular dynamics; and generated neural networks to be trained on data from FHI-aims, a.

Austin, TX, US

Austin, TX, US

First project:Low power two stage fully differential OTA using 180 nm technology in Cadence. First stage was telescopic, second stage common source. The circuit has common mode and differential mode feedback circuits along with frequency compensation techniques. The whole design was based on a single ideal current source.Second projectLow power, low noise.

Champaign, IL, US

Worked on implementing machine learning algorithms to the problem of many body localization. Research available at https://www.ideals.illinois.edu/bitstream/handle/2142/98888/SROP_report.pdf?sequence=2&isAllowed=y

Champaign, IL, US

• Highly parallelized C program for numerical fluid dynamics NumericalFluidDynamics: https://github.com/davidleocadio94/NumericalFluidDynamics
• Earthquake data analysis in the context of modern statistical physics (renormalization group and scaling theories).The Renormalization Group: https://courses.physics.illinois.edu/phys498cmp/sp2022/Ising/RG.html
• Worked on implementing machine learning algorithms to the problem of many body localization.Summer Research Opportunities Program (SROP): https://grad.illinois.edu/diversity/srop

Masters, Electrical And Computer Engineering

Exchange Student, Physics

Bachelor Of Science - Bs, Engineering Physics