Scroll
Multi-agent AI systems and SaaS products engineered to hold up under real-world conditions — not just in demos. This work unites agentic AI design (orchestration, task delegation, and decision-making) with the same performance, security, and reliability engineering applied to correcting inefficiencies in production web systems.
Rakshan Kamalakannan is a B.Tech Information Technology student at Rajalakshmi Engineering College, specializing in multi-agent AI systems and SaaS products engineered to hold up under real-world conditions — not merely in demonstration environments. The work unites agentic AI design (orchestration, task delegation, decision-making) with disciplined performance, security, and reliability engineering.
Agentic AI & SaaS — an agentic AI system is currently under development as part of IBM's SkillsBuild Nextgen program at CSRBOX. Prior experience includes SaaS product development across frontend and backend at Synergech, spanning an agentic chatbot for AI-powered workflow management.
Performance & Reliability — a track record of turning inefficiency into measurable outcomes: up to 30% reduction in page load times • 20%+ improvement in user flow efficiency • strengthened security posture through SSL validation and vulnerability analysis.
An AI agent that's slow or insecure isn't production-ready, no matter how smart its reasoning is. Every project — agent, SaaS feature, or web platform — gets evaluated the same way: does it behave correctly under real load, not just ideal conditions. Currently strengthening foundations in data structures, algorithms, and system design so this scales beyond a prototype.
Designing agentic architectures — orchestration, task delegation, and decision-making across multiple specialized agents — built to work reliably in production, not just in demos.
End-to-end SaaS feature development across frontend and backend, applying the same reliability-first engineering approach used in performance and security work.
Deep-dive analysis of Core Web Vitals, load times, render-blocking resources, and bottlenecks that degrade real-user experience and tank SEO rankings.
End-to-end certificate validation, HTTPS configuration review, and identification of security gaps that erode user trust and flag browser warnings.
Building an agentic AI system as part of IBM's SkillsBuild Nextgen program, applying multi-agent design to enterprise software development, performance optimization, and real-world web systems auditing.
Developed Smart Manager, an agentic chatbot for AI-powered work-from-home and team management — a multi-agent conversational database interface enabling non-technical managers to execute CRUD operations through natural language. Contributions spanned frontend and backend implementation using Angular, FastAPI, LangChain, SQLite, and Redis.
Second-year coursework establishing a strong foundation in information technology, with focused study in data structures, system design, and applied software development, alongside independent multi-agent AI and SaaS project work.
A multi-agent conversational database interface enabling non-technical managers to execute CRUD operations on employee and team data through natural language — eliminating the need for SQL/ORM expertise. Architected as a 5-agent pipeline: Parser Agent (intent parsing), Generator Agent (SQLModel ORM code generation), Critique Agent (code review and self-correction), Loop Evaluator Agent (sandboxed dry-run testing with rollback protection), and Sequence Agent (semantic validation and response formatting). Developed during an engineering internship at Synergech.
A no-code AI-native platform connecting restaurants and food chains with surplus food to NGOs, shelters, and communities in real time. Comprises a multi-agent system — Donor Agent, Matching Agent, Impact Agent, and Conversational Agent — orchestrating end-to-end food redistribution.
A privacy-first AI platform for B2B lead generation with zero API costs — all processing on-device via Ollama. Architected a 3-tier multi-agent system: CEO Orchestrator → Manager Layer (VP Sales, CTO) → Execution Layer agents.
Open to performance audits, web consulting, SaaS collaborations, and high-impact internships. Particularly interested in systems where precision, efficiency, and reliability are critical.