Tushar Torani

Tushar Torani

Full Stack + GenAI Developer | Building RAG, Multi-Agent & AI Applications | React • FastAPI • LLMs

Pune, IndiaArtificial Intelligence and Software Development
1roles
65skills
1education
4credentials

About

I’m a Full Stack Developer and GenAI Engineer passionate about building intelligent products that combine scalable software engineering with modern AI systems.My work spans end-to-end product development—from designing responsive web applications to building production-oriented Generative AI workflows using LLMs, RAG pipelines, multi-agent architectures, and structured data systems.Recently, I completed my internship as a GenAI Intern at Neuleap AI, where I worked on AI applications involving retrieval-augmented generation (RAG), multi-agent orchestration, NL→SQL systems, document intelligence, and full-stack integration.I’ve built projects ranging from a real-time Mental Health Sentiment Analyzer using DistilBERT and FastAPI to agent-based AI systems integrating LangChain, vector databases, and intelligent retrieval pipelines. On the engineering side, I work across React, Node.js, FastAPI, MongoDB, SQL, Supabase, and modern deployment workflows.My focus is turning advanced AI capabilities into reliable, intuitive user experiences—building systems that are not only technically strong but genuinely useful.Core Interests:• Generative AI• Multi-Agent Systems• Retrieval-Augmented Generation (RAG)• Full Stack Development• NLP & Applied ML• AI Product EngineeringCurrently seeking opportunities in Generative AI, Full Stack Engineering, and Applied AI to build impactful products and continue learning at scale.

Experience

GenAI Intern

NeuLeap.AI

5 mos · Pune District

Built and delivered 3+ GenAI and full-stack applications; developed multi-agent workflows and retrieval pipelines integrating LLMs, APIs, and structured databases, improving response relevance by approximately 35%; designed integrations supporting sub-2-second average inference latency; automated document understanding and retrieval workflows, reducing manual effort by approximately 50%. Selected projects included a biomedical RAG and NL-to-SQL multi-agent system processing 1,000+ records and documents, an MCP-based multi-agent workflow system integrating 5+ tools, and an agentic financial 10-K RAG system analyzing 100+ pages of SEC filings.

Education

MIT World Peace University

Bachelor of Technology, Computer Engineering

undefined 2022 - undefined 2026

CGPA: 7.2 / 10.0

Skills

Soft SkillsProblem SolvingGenAI API ExtensionsChatGPTTransformer ModelsWeb InfrastructureDeep LearningDevOpsJavaScript FrameworksLogistic RegressionScikit-LearnMachine LearningExpress.jsWeb IntelligencePC & Mac platformsFront-end EngineeringContext-Aware Conversational AgentsComputer ScienceHuman InterestData-driven Decision MakingFront-End DevelopmentDeveloper ToolsReal-time MonitoringConstructionSoftware DeploymentVisual StudioMongoDBPython (Programming Language)Cascading Style Sheets (CSS)MERN StackGitHubGitJavaScriptNode.jsFastAPIREST APIsBrowserStackPostman APIReact.jsMySQLTailwind CSSFull-Stack DevelopmentE-CommercePythonSQLC++LLMsPrompt EngineeringRAGMulti-Agent SystemsLangChainLangGraphNLPTensorFlowTransformersClassificationSentiment AnalysisHTMLCSSSQLiteVector DatabasesNeo4jPostmanSupabaseVS Code

Projects

Basic Note Webpage for DevTown company

Nothing just a basic project given by the company at the bootcamp yeah but was my first project

Mental Health Sentiment Analyzer

Full-stack ML application for real-time emotion detection in mental health texts using DistilBERT and transformer-based NLP. Achieved 92% accuracy across 6 emotions, enabled real-time inference under 300ms with FastAPI and vanilla JavaScript, and added confidence scoring and emojis.

Multi-Agent AI Customer Support Chat

AI-driven multi-agent chatbot for domain-specific customer support using specialized FAQ, Escalation, and Sentiment Monitor agents. Integrated a RAG pipeline with a vector database, reducing irrelevant responses by 65%, with a targeted 70% reduction in human intervention for routine queries.

Full Stack E-Commerce Website

MERN and Supabase online shopping platform with secure authentication, media storage, payment workflows, product browsing, search, order history, and scalable Node.js and MongoDB APIs supporting up to 10,000 concurrent users.