Isha Sharma

AI/ML and Data Science Engineer with end-to-end production experience spanning model development, LLM fine-tuning, RAG system design, agentic AI deployment, Computer Vision, NLP, and MLOps pipeline management. Delivered 4 production GenAI/ML systems at KPMG India with 60–80% measurable efficiency gains. Strong foundation across TensorFlow, PyTorch, Python, Azure, Docker, and MLflow.

Ranked 79th / 20,000+ teams (top 0.5%) in Amazon ML Challenge 2025. Published researcher in applied GenAI and multi-agent risk systems.

AI/ML Engineering Intern, RIA-COE at KPMG India (2026-01 – 2026-07)

• Architected and deployed 4 production ML and GenAI systems — fraud detection, document intelligence, workflow analytics, and multilingual NLP — using TensorFlow, PyTorch, LLMs, and RAG on Azure AI; achieved 60–80% reduction in processing time across all use cases, validated against defined quality benchmarks.
• Fine-tuned LLM and NLP models using iterative prompt engineering and RAG pipelines across 500+ SQL-indexed policy documents; reduced response time by 70% and retrieval precision by 40% through systematic model evaluation cycles.
• Managed end-to-end ML lifecycle using MLflow for experiment tracking, model versioning, and reproducibility; deployed models on Azure with Docker containerisation and GitHub-based version control; engineered a multi-engine OCR and CV pipeline processing 10,000+ documents, cutting verification time by 60%.
• Deployed a multilingual analytics platform (100+ languages, 3 user tiers) with 80% operational improvement; applied AI governance practices including guardrails, bias checks, and audit-style validation throughout all production deployments.
• Collaborated with 8+ cross-functional stakeholders using Agile workflows to define ML product specifications and drive production delivery on schedule.

B.Tech in Computer Science and Engineering – Lovely Professional University (2022 – 2026)

Data Science and Artificial Intelligence – Daksh Gurukul, IIT Guwahati (2025-01 – 2026-03)