Name : Shyam Sunder Pandey
Role : Full Stack Developer — AI-integrated systems, Cybersecurity, Cloud&DevOps
University : UEM Jaipur — B.Tech Computer Science & Engineering (3rd Year)
Location : Jaipur, India 🇮🇳
Primary Stack: Next.js · React.js · Express.js · FastAPI · Java · Python · C · PostgreSQL · TypeScript
Building : GaonDoc (AI rural health triage) · Agentic AI Security Copilot
Hackathons : AceHack 5.0 — 11th / 500+ teams (IITs, IIITs, NITs) · Laserhacks — International Finalist
Open To : Full Stack · Backend · AI Integration Internshipsshyam-s-space-portfolio.vercel.app
Not a regular portfolio page - a 3D interactive space journey built entirely in the browser. Each section is a themed scene you navigate through, with cinematic transitions between them.each representing a different aspect of my work and identity Cinematic scenetransitions powered by GSAPanimation timelines.
Real-time 3D rendering via Three.js - runs entirely client-side, no plugins needed Responsive across devices built with Next.js 14 App Router + TypeScript + Tailwind CSS
Most developer portfolios are scrollable landing pages. I wanted mine to feel like an experience something that shows frontend engineering depth, not just lists projects. Every scene transition, particle system, and camera movement is hand-coded, not a template.
I'm Shyam — a 3rd year Computer Science student at UEM Jaipur and a full-stack developer who builds AI-integrated systems targeting real infrastructure gaps in India. From a blockchain-based data consent platform aligned to India's DPDP Act 2023, to a voice-first rural health triage system designed for eSanjeevani — I work on problems that matter beyond the hackathon deadline.
I build production-minded systems — not demos. My work spans smart contract development on Algorand, multi-model AI pipelines combining LaMa + ESRGAN + SAM, and zero-cost automated platforms powered by GitHub Actions. I approach every project with one question: what breaks first under real usage?
Highlights: AceHack 5.0 — 11th place among 500+ teams including IITs, IIITs & NITs · Laserhacks 2025 — International Finalist, Lasell University, USA · HackUEM 3.0 — 1st Place
Open To: Full Stack & Backend Engineering Internships · AI-Integrated Product Roles · Open Source Collaboration
I don't ship features — I ship working systems. Every architecture decision I make starts with failure modes, not happy paths. The best code is the code that doesn't need to be explained.
learning:
- System design and distributed systems architecture
- LangGraph multi-agent orchestration for AI pipelines
- Docker + container-based production deployments
building:
- GaonDoc — voice-first AI triage integrated with eSanjeevani API
- Agentic AI Security Copilot — LLM-powered threat detection + response
- Full-stack platforms with FastAPI backends and Next.js frontends
exploring:
- Vector databases for RAG-based AI applications (ChromaDB, Pinecone)
- Advanced PostgreSQL — indexing, query planning, partitioning
- Cloud-native CI/CD pipelines with GitHub Actions + Docker
open_to:
- Full Stack and Backend engineering internships
- AI integration and product engineering roles
- Research-oriented software engineering projectsBuilt and maintained web applications supporting the foundation's social impact programs. Worked across the full frontend lifecycle from UI design to deployment.
- Developed 3+ responsive web interfaces using React and Tailwind CSS with cross-browser compatibility
- Integrated REST APIs to connect frontend dashboards with backend data services
- Improved page load performance by optimizing component rendering and asset delivery
- Collaborated on maintaining and extending the foundation's primary web platform
Stack:
|
I'm a 3rd year CS student, not a lawyer — but India's DPDP Act 2023 created a real engineering problem worth solving. ConsentChain lets users mint ARC-19 NFTs as verifiable consent records, stores encrypted data in an AES-256 vault on IPFS, and uses a Vibekit AI agent to guide non-technical users through the consent flow. The hardest challenge was maintaining consent immutability on Algorand while keeping the UI accessible to non-Web3 users. Tech Stack:
|
|
Four separate AI models. One coherent output. RestorAI chains LaMa (inpainting) → Real-ESRGAN (4× upscaling) → SAM (segmentation masking) → DDColor (colorization) into a single FastAPI-backed pipeline. The engineering challenge was preventing quality degradation at each model handoff — solved by normalizing tensor outputs between pipeline stages before passing to the next model. Tech Stack:
|
|
Most hackathon aggregators charge for listings or go stale within days. HackRadar runs entirely on GitHub Actions as a scheduler — scraping 15+ platforms including Devpost and MLH on a cron, storing results in Supabase, and serving a live Next.js frontend. Zero server cost, zero manual updates. The interesting engineering decision was using GitHub's free compute as a backend cron service instead of paying for infrastructure. Tech Stack:
|
|
India has 1 doctor per 1,445 rural patients. eSanjeevani exists but is underused because of language barriers and low digital literacy. GaonDoc is a voice-first AI triage layer on top of eSanjeevani — patients speak in Hindi or regional languages, the system triages severity, and routes to the right care channel. Building for low-connectivity environments where a 2G connection is the norm, not the edge case. Stack: |
|
Security tools generate noise. This builds signal. A multi-agent LLM system where specialized agents handle threat classification, vulnerability scanning (OWASP-aligned), and automated incident response — with a vector database storing threat intelligence memory across sessions. Goal: reduce mean time to detection for small engineering teams that can't afford a dedicated SOC. Stack: |
| Domain | Level | What I've Built With It |
|---|---|---|
| Computer Vision | Intermediate–Advanced | 4-model restoration pipeline: LaMa + Real-ESRGAN + SAM + DDColor |
| LLM Integration | Intermediate | Vibekit AI agent for ConsentChain guided consent flows |
| Multi-Agent Systems | Learning | LangGraph-based orchestration for Agentic Security Copilot |
| Image Processing | Intermediate | Tensor normalization across model pipeline stages in RestorAI |
| AI-Powered APIs | Intermediate | FastAPI microservices wrapping PyTorch models with structured responses |
| RAG Pipelines | Learning | ChromaDB-backed threat intelligence memory for Security Copilot |
| Recognition | Details | Year |
|---|---|---|
| 🏅 AceHack 5.0 — 11th Place | National hackathon · 200+ teams · Top teams from IITs, IIITs & NITs | 2026 |
| 🌍 International Finalist | Laserhacks 2025 · Lasell University, USA · Competed against global teams | 2025 |
| 🥇 1st Place | HackUEM 3.0 · Internal hackathon · College Tech-related Events (TechUtopia) · UEM Jaipur | 2025 |
I'm actively looking for Full Stack / Backend / AI internship opportunities — especially product-focused teams building things that actually get used.
"Good code is invisible. Great engineering is what makes it stay that way."