Navi Mumbai, India • Multi-Agent AI • Real-Time LLM Systems • AI Infrastructure
I’m a B.Tech IT student from Mumbai focused on building AI systems that are reliable, scalable, and useful in real-world environments. My work revolves around multi-agent systems, real-time LLM pipelines, conversational AI, and understanding how AI can integrate into existing systems effectively.
I love building structured AI systems (not just simple prompt wrappers) and designing production-ready architectures that bridge advanced research and tangible software solutions.
Professional Experience
- •Optimized database query performance and refactored relational database models using PostgreSQL and Prisma, leading to 35% faster page load times.
- •Designed and implemented clean RESTful API endpoints for user onboarding workflows, dashboard metrics, and document search indexing.
- •Worked closely with front-end teams to integrate responsive, pixel-perfect layouts using React, Tailwind CSS, and Next.js.
- •Maintained unit and integration test coverage across server routes, ensuring zero regressions on production deployments.
Next.js
React
TypeScript
Node.js
PostgreSQL
Prisma
Tailwind CSS- •Engineered low-latency WebSocket connection handlers for audio streaming to integrate real-time voice with LLMs, reducing processing delays to < 200ms.
- •Developed reliable fallback systems and custom buffer sizing for audio chunks, ensuring steady voice interaction under fluctuating network conditions.
- •Integrated text-to-speech (TTS) and speech-to-text (STT) services (Deepgram, Cartesia, ElevenLabs) with custom caching layers, speeding up generation times.
- •Configured multi-agent orchestrations for system routing based on semantic classification of user audio input, enabling dynamic conversational flows.
Python
FastAPI
WebSockets
Redis
Docker- •Architected a responsive, role-based attendance management system for 150+ employees, automating data entry and reducing administrative overhead by 40%.
- •Developed real-time analytics dashboards to visualize workforce metrics, enhancing managerial decision-making efficiency by 30%.
- •Optimized database query performance and backend logic, resulting in a 25% reduction in page load latency.
JavaScriptPostgreSQLNode.jsReactProjects
Research Tracks
AI-Driven Digital Twins
Behavioral Simulators & Industrial Logic
Modeling complex software and physical architectures using AI agents. Investigating how simulation environments can recreate human-like interaction loops, enabling automated safety checks, state drift analysis, and stress-testing of software constraints.
AI Safety, Evaluation & Alignment
Multi-Agent Consensus & Guardrails
Developing robust evaluation methodologies to assess multi-agent reliability and alignment. Researching consensus algorithms in decentralized agent networks, real-time prompt-drift detectors, and active self-evaluation guardrails to eliminate hallucination risks.
Education & Achievements
Education
Radcliffe School
Completed 12th grade
KJ Somaiya School of Engineering
B.Tech in Information Technology (Honours in Artificial Intelligence)
Achievements & Awards
Gemini HackDays
Google Developer Groups / Firebase
ETH Mumbai
Devfolio / Ethereum India
GDSC Hackathon
Google Developer Student Clubs
NPCI BharatAI (GFF 2025)
National Payments Corporation of India
Tech Stack
GitHub Contributions
Leadership & Activities
Tech Head, DataZen
KJ Somaiya School of Engineering
Spearheaded technical execution for AI workshops and hackathons, mentoring 300+ students in Full Stack and ML concepts. Led technical event strategy, managed speaker sessions, and oversaw hands-on bootcamps on RAG architectures and multi-agent systems.
Creative Head, SMLRA
KJ Somaiya School of Engineering
Directed branding strategies and managed cross-functional teams to execute campus-wide technical events. Oversaw visual identity, promotional campaigns, and coordinated logistics for large-scale college events attended by hundreds of students.