Open to opportunities

Ayush Tiwari

@ayushtiwari5

Message

ML Research Engineer specializing in LLMs, RAG, and production ML.

India

Message

What I'm looking for

I seek roles building production ML and LLM systems—focusing on RAG, multi-agent AI, model optimization, and scalable deployment within collaborative, research-driven teams.

I am an ML Research Engineer focused on large language models, generative AI, and production-ready ML systems. I specialize in fine-tuning LLMs, building retrieval-augmented generation (RAG) pipelines, and optimizing models for low-latency deployment.

My work includes architecting multi-agent Agentic AI systems, improving inference and orchestration for complex workflows, and applying state-of-the-art techniques across NLP and computer vision. I have hands-on experience with PyTorch, Transformers, diffusion models, and model optimization strategies such as QLoRA and LCM.

I bring measurable impact from internships and industry roles — reducing latency, improving model accuracy, and optimizing cloud costs — and I enjoy tackling challenging research-to-production problems that blend deep learning research with scalable engineering.

Experience

Work history, roles, and key accomplishments

Current

ML Research Engineer

Current

Gupshup

Mar 2024 - Present (2 years 3 months)

Optimized legacy read-prediction code reducing latency by 80% and architected/deployed multi-agent Agentic AI systems with dynamic routing and context sharing for complex RAG workflows.

PyTorch Transformers RAG Agentic Systems Optimization MCP

ML Dev Intern

OpenAI

Sep 2022 - Nov 2022 (2 months)

Improved medical image classification accuracy by 1.7% using deformed attention and contributed to large-scale generative model data pipelines focused on quality and efficiency.

Deformed Attention Data Pipelines Generative Models Data Quality Model Evaluation PyTorch

Cloud Consultant Intern

Microsoft

Jan 2022 - Jul 2022 (6 months)

Delivered FinOps and Dynamics 365 solutions and optimized Azure infrastructure for The Metropolitan Museum of Art, reducing cloud costs by 30% via resource allocation and automation.

Azure FinOps Cloud Optimization Dynamics 365 Cost Management Automation