Open to opportunities

ashwathi pillai

@ashwathipillai1

Message

AI Operations Specialist improving LLM groundedness with RAG/GRAG prompts and rigorous evaluation.

India

Message

What I'm looking for

I’m looking for junior–mid roles where I can design RAG/GRAG prompts, build evaluation/reviewer systems, and improve production groundedness with rigorous A/B testing. I want cross-functional teams and clear ownership from UAT to launch.

I’m an AI Operations Specialist with 2 years of hands-on experience in LLM prompting, RAG/GRAG prompt design, LLM evaluation, knowledge-base structuring, and model performance testing. I’ve consistently improved chatbot groundedness by 30–50% across 5+ multi-industry deployments (education, finance, healthcare, telephony).

I build reviewer prompts, groundedness evaluators, simulators, and reusable prompt frameworks that turn evaluation into an engineering workflow. I design expert evaluator logic that classifies responses as grounded, partially grounded, or ungrounded using structured JSON—supporting regression testing and model comparison.

In production-focused work, I run multi-model testing across OpenAI, Gemini, and AI Studio, perform continuous A/B prompt testing, and debug LLM/RAG/GRAG issues with root-cause analysis. I’m currently upskilling in LangChain basics and advanced RAG architectures to deepen my technical depth for engineering-track roles.

Experience

Work history, roles, and key accomplishments

Current

AI Operations Specialist

Current

Mantra Digital

Jun 2024 - Present (2 years)

Designed RAG/GRAG prompts and simulators that automated groundedness checks and improved chatbot groundedness by ~30–50% across 5+ deployments. Built JSON-based groundedness evaluators and conversation-reviewer prompts to detect instruction violations and intent mismatches, reducing false positives through iterative A/B prompt testing and multi-model evaluation.

RAG GRAG Prompt Engineering LLM Groundedness Evaluation Prompt Testing Retrieval Accuracy Testing Multi Model Testing (OpenAI Gemini Claude Instruction Intent Violation Detection Regression Testing

Pharmacovigilance Intern

Pristyn Research Solutions

Jun 2021 - Nov 2021 (5 months)

Completed clinical data management and medical writing work, including ICSR development and case narrative writing with a focus on structured, accurate documentation.

Clinical Data Management Medical Writing ICSR Development Narrative Writing

Communication Trainer

Independent

Trained 50+ students in structured communication and clarity, applying coaching techniques to conversational AI design such as chatbot tone calibration and intent structuring.

Communications Coaching Clarity Improvement Tone Calibration