Ayush Tiwari

I’m a Python Engineer focused on LLM engineering—turning agentic ideas into reliable, production-grade systems. I studied Computer Science & Engineering (B.Tech) and built a strong foundation in software and data-driven problem solving.

As a Python Engineer - LLM, I led a cross-functional pod delivering end-to-end LLM pipelines and agentic backend services. I designed multi-agent reasoning workflows (vision + text inputs), built adversarial prompt evaluation pipelines to expose failure modes, and improved alignment using SFT and RLHF-style output reviews.

I also build practical systems beyond pipelines: Mnemosyne, an offline agentic multimodal memory system with OCR + embeddings and hybrid retrieval (LanceDB), plus deployed REST services like a Flask-based heart disease prediction API and a real-time MERN chat app. I care about robustness, evaluation, and clean delivery—so models perform safely, consistently, and at scale.