Ajoy Debnath

I’m a Computer Science & Engineering B.Tech student who builds production-ready full-stack web applications with real-time features and LLM-powered capabilities. I care about system design, measurable performance, and reliable delivery—from API design and authentication to monitoring and iteration. My goal is to ship products that feel fast, intelligent, and dependable in real-world usage.

As an AI Systems Developer Intern at Mindsparc Innovation, I engineered LLM-powered log correlation pipelines ingesting 10K+ events/day. I reduced root-cause analysis time by 60% using anomaly detection, contextual grouping, and automated alert summarization. I also built async REST APIs with job queues for 500+ daily tasks, improving availability and cutting processing latency by 45% through retry logic, health checks, and circuit breakers.

Freelancing as a Full Stack Developer, I’ve delivered three production apps—an EdTech LMS, a patient booking portal, and a social platform—using JWT auth and RBAC while optimizing schemas for 5,000+ monthly requests. I owned the full delivery lifecycle, including system design, REST APIs, third-party integrations like Razorpay and OAuth 2.0, and AWS/Vercel deployment, with post-launch monitoring via Sentry.

In projects, I combined modern frontend, backend, and AI tooling to produce measurable results: OfferPilot’s ATS resume scoring runs with sub-10ms latency and its APIs were load-tested to 300 RPS using k6. Medox AI delivered an OpenAI-powered symptom triage engine (~85% accuracy) and CampusConnect used AWS Rekognition and OCR identity checks plus a real-time match & chat engine for 300+ concurrent users at sub-80ms latency.