Third-year B.Tech Computer Science student at Vellore Institute of Technology, Bhopal (Class of 2027), currently balancing three concurrent engineering internships alongside coursework.
My work sits at the intersection of three disciplines that are usually treated separately: agentic AI systems that orchestrate themselves, data engineering that holds up at government scale, and systems-level programming that respects the machine it runs on. Recent work includes a seven-agent compliance engine that evaluates policy documents against six ISO standards in a single pass, ML pipelines that reduced preprocessing latency by 35% on large-scale citizen datasets, and a desktop utility that suspends live OS processes and restores exact workspace state.
| Execution | AI/ML Impact | Systems |
|---|---|---|
| 3 concurrent internships MPOnline · ServiceNow · Campus Bites |
87%+ model accuracy citizen-request triage |
7-agent orchestrated pipeline SentriX |
| 6+ shipped projects 2 live in production |
35% latency reduction data preprocessing |
OS-level process suspension Stash |
| Project | Description | Stack |
|---|---|---|
| SentriX | Agentic AI compliance-assessment platform. A seven-agent pipeline scores policy documents against six ISO standards, backed by a three-tier hybrid scoring engine (ML semantic → Groq → NLP fallback) with real-time SSE progress streaming. | TypeScript React Express Claude Groq |
| SageLink | Three-role eldercare platform (Elder / Caregiver / Clinician) featuring a Gemini-powered memory companion, TensorFlow.js face recognition, cognitive-decline pattern detection, and real-time caregiver alerting across an 11-table schema. | React TypeScript Supabase Gemini |
| Secure-OSINT-FaceID | Privacy-first smart doorbell: 5 FPS local face matching (MediaPipe + DeepFace/ArcFace) with 3D liveness anti-spoofing and a human-gated OSINT layer that requires explicit operator approval before any public lookup. | Python FastAPI MediaPipe DeepFace |
| Stash | Desktop resource-optimization utility that suspends running applications at the OS process level to reclaim CPU, RAM, and battery, then restores the exact workspace state losslessly. | Python OS Process Management |
| UrbanGrocer | Full-stack grocery delivery platform with atomic checkout-state handling, indexed sub-100ms product lookups, and asynchronous cart management. Live ↗ | React Node.js MongoDB |
| FitPal | Production-deployed MERN fitness tracker with JWT authentication, progress analytics, and nutrition tooling. Live ↗ | MERN JWT GSAP |
Campus Bites Full-Stack Development Intern / Tech Lead Jul 2026 – Present
Cross-functional team leadership; zero-error transactional order processing
ServiceNow Virtual Platform Engineering Intern May 2026 – Jun 2026
Flow Designer automation, ATF regression suites, SLA analytics
ServiceNow Micro Certification
MPOnline (Govt. MP) AI/ML & Software Engineering Intern Mar 2026 – Present
35% faster data pipelines · 87%+ triage model accuracy · MVC/REST at scale
- Agentic AI workflows — multi-agent orchestration patterns beyond single-LLM calls
- LLM fine-tuning — domain-specific model optimization
- Advanced Next.js — server components and edge computing
Hackathon teams in agentic AI · open-source contributions · research collaboration in computer vision and multi-agent systems · mentoring junior developers

