Ashish Singh

• Built a multi-agent system using Google ADK to automate SOW vs. client transcript comparison, generating gap
analysis, CTF insights, and scope review using Vertex AI Search
• Developed downstream agents to transform gap analysis outputs into solution design documents and
client-specific Excel templates with metadata and structured instructions
• Deployed agents on Vertex AI Agent Engine with integration into Gemini Enterprise, leveraging Googles Gemini
and Anthropics Claude for advanced reasoning and document processing

Intelligent Agentspace RAG based
• Developed AI agent using Googles Agent Development Kit (ADK) to develop a custom agent that provides
detailed solution based on past jira tickets, web search and Gemini LLM
• Leveraged GCPs cloud function, scheduler, Vertex AIsearch datastore, Gemini, secret manager to create a RAG
system
• Deployed ADK agent into agent engine and registered it in AgentSpace to give a seamless experience to the end
user

LLM powered Business Metrics Analyser:
• Developed architecture for executing an end to end GenAI solution for a bank
• Leveraged Gemini 1.5 pro to ingest BQ data and generate commentary and identify trends
• and insights from several Business metrics
• Deployed

Heart Disease Detection Classifier:
• Developed a classifier using various algorithms (LR, KNN, RF) to predict heart disease from
clinical data, employed data visualization for insightful analysis and hyperparameter tuning to
achieve remarkable accuracy of 86%.

SQL query pattern analysis:
• Developed NLP-based SQL query solution Doc2Vec embeddings, t-SNE
for dimensionality reduction, ANN-based clustering (ANNOY), identified GB
queries cosine similarity and provided to SMEs for optimization.

SQL Query Generation AI:
• solution PaLM LLM SQL queries natural databases, RAG pipeline