Nandini Bajaj

AI Engineer Intern

Sargvision

June 2026 – Present

Building agentic AI features end-to-end on Google ADK and Vertex AI — LLM pipelines, agent-to-agent workflows, real-time voice, and the deployment path around them.

• •Engineered an LLM-powered blog generation pipeline automating content production end-to-end, from source ingestion through refinement to publishable draft.
• •Implemented agent-to-agent (A2A) workflows and custom agent skills, enabling task handoff between agents and extending built-in ADK capability.
• •Integrated Vertex AI’s Gemini Live model for real-time speech-to-speech voice conversation.
• •Contributed to the platform’s GCP deployment (Cloud Run, Firestore) and GitHub Actions CI/CD pipeline with an automated agent-evaluation gate.
• •Built and refined UI components to surface, control, and review AI-generated content within the product.

AI Intern

IBM (Project-Based)

Sep 2025 – Oct 2025

Engineered a Python-based NLP and machine learning application end-to-end, achieving reliable real-time classification in production-style usage.

• •Owned the full modelling loop — problem decomposition, feature and data design, and model evaluation — to achieve reliable real-time classification.
• •Designed data preprocessing, validation, and ETL pipelines feeding the model.
• •Deployed model predictions through FastAPI RESTful microservices, applying systematic debugging to isolate and resolve reliability issues under real usage.