Open to opportunities

pasala pravalli

@pasalapravalli

Message

ML Engineer · 3 deployed AI apps on HuggingFace Spaces · Production RAG, LLM pipelines & multi-agent systems

India

Message

What I'm looking for

I'm looking for remote roles where I can build production LLM applications — RAG and multi-agent systems — with fast latency, strong evaluation, and cost-effective deployment, ideally using Python, FastAPI, LangChain, and PyTorch. I work async-first and am flexible across time zones.

I'm a Machine Learning Engineer with 3 years at PwC building production data and AI systems for Fortune 500 clients including BioMarin, Kimberly Clark, and Rockwell. I architected data validation pipelines processing 50TB+ of data, ensuring ML-ready formats and reducing manual validation time by 40%.

Independently, I've built and deployed 3 live AI projects: a RAG chatbot powered by Groq's LLaMA 3.3-70b with multi-PDF support and conversation history, an LLM Comparison Dashboard benchmarking models with Gradio and SQLite, and SafeRoute AI — all live on HuggingFace Spaces and GitHub.

I optimized RAG retrieval from 10s to sub-second through chunking strategy, semantic search tuning, and re-ranking. I love closing the loop between research and production — turning LLM capabilities into fast, cost-effective systems real users interact with.

Experience

Work history, roles, and key accomplishments

Current

Data & AI ML Engineer

Current

PwC

Mar 2023 - Present (3 years 3 months)

Architected data validation pipelines processing 50TB+ for Fortune 500 clients (BioMarin, Kimberly Clark, Rockwell), ensuring ML-ready data formats. Automated reconciliation reporting with Alteryx to reduce manual validation time by 40% and built anomaly detection workflows improving data quality issue detection by 25%.