Ravinarayana A
LLaMA Post-Training for Agentic AI at Meta | Ex-Applied Scientist at Amazon Rufus | LLM | Deep Learning | NLP | Large Scale Deep Learning
- Role
- Research Engineer at Meta
- Location
- San Francisco, CA, US
- LinkedIn followers
- 500 followers
About Ravinarayana A
As an accomplished Applied Scientist with extensive experience at industry giants like Amazon, Comcast, and Sony, I excel in leveraging cutting-edge Large Language Models (LLMs) and advanced machine learning techniques to drive innovative solutions and elevate customer experiences. My expertise spans various domains, including large language modeling, representation learning, and large-scale distributed training.At Amazon, I lead product recommendation experience for Rufus, an AI-powered shopping assistant Additionally, I ideated and implemented a separate solution for generating shopping concepts using LLMs for broad query searches on Amazon, resulting in an impressive $30 million annual revenue uplift. I have designed and built large transformer-based representation models through distributed training setups, utilizing parallelized frameworks and high-performance computing infrastructure, contributing to a 420 basis points precision improvement for the Amazon Fashion team.My propensity models for targeted ads in Amazon Payments, utilizing billions of payment transaction data points, have driven a substantial $1 million annual revenue uplift. Furthermore, I have investigated and applied methods for balancing losses in very large multi-task learning settings, optimizing model performance across multiple objectives simultaneously.As a Senior Software Engineer at Sony India Software Centre, I led projects that generated approximately 10 million JPY in revenue by automating comparison processes between pre-production and production websites. Notably, I showcased a suite of AI-based products like an Open Question Answering System built using Spacy at the Sony Technology Exchange Fair in Tokyo, demonstrating potential cost reductions in areas like customer service, HR, and clerical-related processes. Additionally, I presented a novel NLP suite of products, including features like summarization and a contract management system, at the Sony AI Conference in Tokyo.Technical Skills:ML/DL: Large Language Models (LLMs), Generative AI, Distributed Training, Multitask Learning, Multimodal Learning, Transformer Models, NLP (Natural Language Processing)Technologies: Representation Learning, Large Scale Distributed Training, RAG, Instruction Fine TuningTools: PyTorch, DeepSpeed, PySpark
Experience
Research Engineer
Dec 2024 — Present · Burlingame, CA, US
Post-training LLaMA models for wearable devices to enable agentic capabilities and optimize wearable-specific skills.
Education
The National Institute Of Engineering, Mysore
Bachelor of Engineering (B.E.), Information Science
University of Southern California
Master of Science - MS, Computer Science
Skills
- C#
- Web Applications
- Data Structures
- Scipy
- Artificial Intelligence
- Sql
- Linux
- Javascript
- Web Development
- C++
- C
- Programming
- Git
- Php
- Natural Language Processing
- Shell Scripting
- Xml
- Android Development
- Databases
- Neural Networks
- Django
- Data Mining
- Numpy
- Object-Oriented Programming (Oop)
- Java
- Recommender Systems
- Machine Learning
- Mysql
- Software Engineering
- Algorithms
- Html
- Python
- Bootstrap
- Windows
Find verified contacts for anyone on LinkedIn
Unifers gives sales teams verified emails and direct dials, enriched profiles, and outreach that lands in the inbox.
Free plan included · No credit card required
This profile is compiled from publicly available professional sources. Unifers is not affiliated with or endorsed by LinkedIn. Request removal of this profile.