Robert Jr. R.

Diffusion, LLMs, etc… • Gemini 2.0, 2.5, 3.0 • Nano Banana Pro • World Sim • Veo • Imagen 2/3/4

• Trained Large Language Models (LLMs) for code generation and translation • Developed internal training infrastructure for LLMs from scratch, built and trained Splunk AI Assistant https://www.splunk.com/en_us/blog/it/building-an-ai-assistant-for-splunk.html • Designed and developed cloud & on-prem services that power all of Splunk's Experiences Platform

• Large scale cloud messaging service interfacing with K8s, DynamoDB, Redis, and Memcached

• Trained and deployed neural networks for estimating bond prices that contain embedded options while utilizing real-time market data to service 200 million+ requests a day

• Led state of the art GAN research focused on WGANs for anomaly detection on market securities

• Designed and trained Siamese neural network for copyright image violation tool

• Implemented Knowledge Graph in Neo4j to enhance search engine quality • Achieved 70% accuracy predicting required work-hours to complete various business objectives using SVMs, decision trees and ensembling methods • Consulted for a FinTech application in regards to NLU and predictive methods

• Led development of levitation system & placed top 15/150 globally in Hyperloop One's Global Competition • Featured in: Time & TechCrunch - https://time.com/4207248/hyperloop-pod-design-competition/ - https://techcrunch.com/2016/11/02/waterloo-wows-with-demo-of-first-working-hyperloop-air-levitation-system/

• Quantified over 100 days worth of data efficiently in Maestro for the PICO Experiment under the direction of Professor Nikolai Starinski in hopes of being the first team in the world to detect Dark Matter. • Applied bubble chamber techniques to calibrate a gamma spectrograph for the radioactive isotope CS 137.