Vishnuprasath User

I’m a Google Certified Professional Machine Learning Engineer focused on LLM training and applied AI infrastructure. I build end-to-end pipelines—from ETL and statistical modeling to scalable deployment—so models perform reliably in real-world workflows.

As a Delivery Data Analyst | LLM Trainer at Turing (remote), I engineered ETL pipelines for RLHF and SFT, improving data cleaning, deduplication, and preference-pair formatting. I also built low-latency RAG systems using semantic chunking, dense vector indexing, and re-ranking to query 100k+ token corpora. To speed training data generation, I replaced manual labeling with Vision-LLMs and engineered agentic datasets from screen-recorded demonstrations.

I’m equally hands-on with alignment and evaluation: I designed adversarial SFT prompts to stress-test multimodal reasoning and surface edge-case failure modes. As a Freelance AI Trainer (Scale AI), I trained models on data labeling plus STEM and linguistic tasks—performing text ranking, classification, preference evaluation, and instruction follow-up assessments—while maintaining high-quality standards for fine-tuning pipelines.