|
2nd-year B.Tech (AI & ML) @ St. Joseph's College of Engineering, Chennai 🇮🇳 I am a passionate AI engineer with hands-on experience building intelligent systems and automating data workflows. My work sits at the intersection of research-grade AI and production-ready engineering — from emotionally-aware multimodal agents to real-time computer vision proctoring tools. Core Focus Areas:
|
|
# ── vyas.config.yaml ──────────────────────────────────────────────────────────
identity:
name : "Vyas S"
role : "AI Engineer & Full-Stack Developer"
degree : "B.Tech — Artificial Intelligence & Machine Learning"
college : "St. Joseph's College of Engineering, Chennai"
year : 2
cgpa : 7.51
location : "Chennai, India 🇮🇳"
status : "🟢 Open to Internships & AI Collabs"
current_internship:
company : "IQmath Technologies"
role : "AI Engineer Intern"
mode : "Hybrid"
duration : "Jun 2026 – Jul 2026"
focus : ["Generative AI", "ML pipelines", "Real-world AI application development"]
past_internships:
- company : "Stackrack Tech India Pvt Ltd"
role : "React & AI Developer Support Intern"
period : "Dec 2025 – Present"
- company : "Inspire Softech"
role : "Artificial Intelligence Intern"
period : "Jun 2025 – Jul 2025"
ai_stack:
llm_frameworks : ["LangChain", "LlamaIndex", "OpenAI API", "Anthropic Claude", "Google Gemini"]
ml_libraries : ["PyTorch", "TensorFlow", "Keras", "scikit-learn"]
vision : ["YOLOv8", "OpenCV", "MediaPipe"]
vector_dbs : ["Pinecone", "ChromaDB", "FAISS", "MongoDB Vector Search"]
deployment : ["FastAPI", "Docker (basic)", "Vercel", "Render", "GitHub Pages"]
flagship_projects:
- name : "Nova AI"
desc : "Multimodal Emotional AI — vision encoder + audio encoder + text fusion, SLM psychological analysis"
stack : ["Python", "FastAPI", "React 19", "TensorFlow", "LangChain", "Google Gemini"]
status : "🟢 Active"
- name : "ExamGuard Pro"
desc : "AI proctoring — YOLOv8, MediaPipe, Tesseract OCR, dynamic risk scoring, Chrome Extension"
stack : ["Python", "FastAPI", "React 19", "YOLOv8", "Supabase"]
status : "🟢 Active"
- name : "StudyBuddyPro"
desc : "Adaptive AI study suite — offline smart quizzes, flashcards, gamified Pomodoro, RAG pipeline"
stack : ["Python", "React", "Node.js", "SQLite", "Pinecone"]
status : "🔨 WIP"
- name : "NEXUS"
desc : "Multimodal RAG system — ChromaDB + NetworkX semantic graph, Groq→Gemini→Ollama LLM fallback"
stack : ["Python", "ChromaDB", "NetworkX", "Celery", "FastAPI"]
status : "🟢 Active"
community:
- "GDG Chennai — Member"
- "GDG on Campus SJCE — Member"
- "CodeHack Pirates — Community Lead"
currently_learning:
- "Agentic AI workflows (AutoGen, CrewAI)"
- "Retrieval-Augmented Generation (RAG) fine-tuning"
- "Multimodal LLMs (vision + language)"
- "Efficient inference & model quantization"
fun_fact: >
I once shipped a working LLM agent, a REST API backend, a Next.js frontend,
AND a pitch deck — all in a single 24hr hackathon. Red Bull was involved. 🎤
# ──────────────────────────────────────────────────────────────────────────────
🔥 Nova AIMultimodal Emotional Intelligence System Combines facial expression recognition (Vision Encoder), vocal tone analysis (Audio Encoder), and text into a unified emotional context vector. A specialized SLM layer generates session-level psychological well-being reports with stress profiling and suggested interventions. Key features: Multimodal fusion · SLM psychological engine · Living memory · Adaptive UI · Privacy-first local storage
|
AI-Powered Real-Time Exam Proctoring Multi-modal proctoring with MediaPipe face detection, gaze vector tracking, YOLOv8 unauthorized device detection, and Tesseract OCR for forbidden-content screening. Dynamic weighted risk scoring (Safe / Review / Suspicious) with a WebSocket live dashboard and automated PDF violation reports. Key features: 4-layer cheat-resistance · Chrome Manifest V3 extension · JWT multi-agent pipeline · Real-time Supabase DB
|
|
Adaptive AI-Powered Learning Suite Offline-capable smart quiz and flashcard generation from student notes using local NLP algorithms. Spaced repetition engine adjusts difficulty in real time. Gamified Pomodoro timer with XP/coin rewards, daily streak tracking, and JWT-auth local-first SQLite architecture. Key features: Offline AI · RAG pipeline · Spaced repetition · Gamification · Local-first SQLite
|
Multimodal RAG System Dual database engine combining ChromaDB vector search with a NetworkX semantic knowledge graph. Features JWT/RBAC auth, a Celery ingestion pipeline, and a Groq → Gemini → Ollama LLM fallback chain for resilient, production-grade document intelligence. Key features: Dual DB engine · Semantic graph · LLM fallback chain · RBAC auth · Celery pipeline
|
-2496ED?style=for-the-badge&logo=docker&logoColor=white){kind=icon}
-2088FF?style=for-the-badge&logo=github-actions&logoColor=white){kind=icon}
| Role | Company | Period | Mode |
|---|---|---|---|
| 🤖 AI Engineer Intern | IQmath Technologies | Jun 2026 – Jul 2026 | Hybrid |
| ⚛️ React & AI Developer Support Intern | Stackrack Tech India Pvt Ltd | Dec 2025 – Present | Remote/On-site |
| 🧠 Artificial Intelligence Intern | Inspire Softech | Jun 2025 – Jul 2025 | Remote/On-site |
| Certification | Issuer | Year |
|---|---|---|
| Google Cloud Vertex AI Prompt Design | 2025 | |
| Artificial Intelligence Fundamentals | IBM | 2025 |
| Deep Learning & Deep Learning using TensorFlow | IBM | 2025 |
| Building AI-Powered Search with MongoDB Vector Search | MongoDB | Jun 2025 |
| Introduction to Data Science | Cisco | May 2025 |
| Data Science Foundation | Infosys Springboard | Sep 2025 |
| Tata Group — GenAI Powered Data Analytics Job Simulation | Forage | Jun 2025 |
| Oracle Database Administration | Oracle / IIT Bombay Spoken Tutorial | 2025 |
| Role | Organisation |
|---|---|
| 👥 Member | Google Developer Group (GDG) Chennai |
| 👥 Member | GDG on Campus — St. Joseph's College of Engineering |
| 🚀 Community Lead | CodeHack Pirates |
Trophies are powered by github-profile-trophy — if they don't load locally, they will render correctly on GitHub.
⚙️ Set up the Snake for YOUR contributions
Create .github/workflows/snake.yml:
name: Generate Snake 🐍
on:
schedule:
- cron: "0 0 * * *"
workflow_dispatch:
jobs:
generate:
runs-on: ubuntu-latest
steps:
- uses: Platane/snk@v3
with:
github_user_name: veeda241
outputs: |
dist/github-contribution-grid-snake.svg
dist/github-contribution-grid-snake-dark.svg?palette=github-dark
- uses: crazy-max/ghaction-github-pages@v3
with:
target_branch: output
build_dir: dist
env:
GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}Then run it once manually from the Actions tab.
💡 I'm always open to AI research collabs, hackathon teams, internship opportunities, and anything that ships fast and solves real problems.
⭐ If a project made you think "that's clever" — you know what to do.