🚀 I'm a CS sophomore at VIT Amaravati specializing in AI & ML, passionate about building intelligent systems that actually ship. From LangGraph-orchestrated prediction pipelines and autonomous AI agents to ESP32-powered smart glasses and fully local privacy-first AI systems — I build end-to-end, from silicon to software.
|
🏅 VIT Internal Expo Smart Vision Aid Selected |
🎓 CGPA 8.35 B.Tech CS · AI & ML |
🤖 7 AI Projects Shipped & Deployed |
📈 LangGraph 8-Node AI Pipelines |
📍 VIT Amaravati Class of 2028 |
class AnkitSengupta:
def __init__(self):
self.name = "Ankit Sengupta"
self.role = "AI Engineer & Backend Developer"
self.university = "VIT Amaravati · Class of 2028"
self.cgpa = 8.35
self.skills = {
"AI / ML" : ["RAG Pipelines", "AI Agents", "LangGraph", "LangChain",
"TensorFlow", "YOLO", "OpenCV", "XGBoost", "RL Feedback"],
"LLM Infra" : ["LangSmith", "ChromaDB", "FAISS", "Qdrant", "HuggingFace",
"Gemma-3", "Phi-3", "Prompt Engineering", "Vector Embeddings"],
"Backend" : ["Django", "FastAPI", "Flask", "Node.js",
"MongoDB", "WebSockets", "APScheduler"],
"Languages" : ["Python", "JavaScript", "TypeScript", "Java", "C++", "Dart", "MATLAB"],
"Mobile" : ["Flutter", "Android", "Kotlin", "Firebase"],
"Embedded" : ["ESP32", "Arduino", "C++", "IoT Sensors"],
}
self.trophy = "🏅 Selected · VIT Internal Expo 2025"
self.superpower = "ML theory → production systems"
self.status = "🟢 Open to internships & collaborations!"
def say_hi(self):
print("Let's build something that matters 🚀")── 🤖 AI & Intelligence ──
── 🔗 LLM Orchestration & RAG ──
── 🌐 Backend & Web ──
── 💻 Languages & Frameworks ──
── 📱 Mobile & Embedded ──
🔥 "Every project is a chance to bridge theory and real-world impact"
Fully local, privacy-first personal intelligence system — query your entire life: documents, videos, audio, and code using natural language. No data leaves your machine.
-00D4AA?style=for-the-badge&logo=langchain&logoColor=white){kind=icon}
| 🖥️ Frontend | Flutter desktop UI — cross-platform, native-speed, replacing PyQt6 |
| 🧠 LLM Core | Phi-3 Mini (4-bit quantized, fully local) — no internet, no leaks |
| 📦 Vector DB | Qdrant — multi-modal collections: text, PDF, audio, video, code, image |
| 🔍 Hybrid Search | Vector semantic search + BM25 keyword ranking with description boost (1.25×) |
| 🎙️ Media Pipeline | Whisper transcription → chunked embedding → timestamp-precise retrieval |
| 🔗 Orchestration | Migrating to LangChain + LangGraph for modular, graph-based RAG pipelines |
| 🌐 Network | LAN device discovery + SMB remote ingest — index your whole home network |
| 🔒 Privacy | SHA-256 deduplication · All inference local · Zero cloud dependency |
+─────────────────────────────────────────────────────────────────+
│ CYBORG AI ARCHITECTURE │
+─────────────────────────────────────────────────────────────────+
│ │
│ INPUT (docs / video / audio / code / images) │
│ ↓ │
│ INGEST: SHA-256 dedup → extract → chunk → embed │
│ ↓ │
│ STORE: Qdrant (vector) + BM25 (keyword) │
│ ↓ │
│ RETRIEVE: Hybrid search → description boost → top-k │
│ ↓ │
│ GENERATE: Phi-3 Mini (local 4-bit) → streaming response │
│ ↓ │
│ Flutter UI ← LangGraph Orchestration (migration in progress) │
│ │
+─────────────────────────────────────────────────────────────────+
🔄 Migration Status: Moving from standalone Python scripts → LangChain + LangGraph orchestrated pipeline · UI rebuilt in Flutter (cross-platform, replacing PyQt6)
AI-powered stock direction prediction using real-time news, LLM reasoning, RAG pipelines, LangGraph orchestration, and reinforcement learning feedback
| 🔗 Orchestration | 8-node LangGraph pipeline: fetch → sentiment → embed → stock → RAG → LLM → ML → ensemble |
| 🧠 LLM Core | Gemma-3-4B-IT (local HuggingFace) for news reasoning & directional prediction |
| 📦 RAG Layer | ChromaDB / FAISS vector store + sentence-transformers all-MiniLM-L6-v2 embeddings |
| 🤖 ML Model | XGBoost binary direction classifier — combines with LLM for ensemble final call |
| 🔁 RL Feedback | Custom reward loop: record → resolve → retrain XGBoost on real outcomes after 3 days |
| 🌐 Backend | FastAPI + WebSockets (real-time pipeline streaming) + APScheduler background jobs |
| 📊 Frontend | React + Vite + TailwindCSS trading UI — candlestick charts, live price feed, sentiment gauge |
| 🔍 Observability | LangSmith tracing for full LangGraph pipeline visibility |
+------------------------------------------------------------------+
| NEWS SOURCES: NewsAPI · GNews · AlphaVantage |
+----------------------------+-------------------------------------+
|
v
| LANGGRAPH: fetch_news → analyse_sentiment → embed_store |
| → fetch_stock → rag_retrieve → llm_reason |
| → ml_predict → final_predict |
| [ Gemma-3-4B LLM ] + [ ChromaDB RAG ] + [ XGBoost ] |
|
+--------------+--------------+
v v
FastAPI + WebSocket React Trading UI
|
v
RL FEEDBACK LOOP → Record → Resolve → Reward → Retrain
Real-time AI-powered assistive smart glasses — giving spatial freedom to the visually impaired
| 📷 Vision System | 3× ESP32-CAM modules + ultrasonic sensors → full 360° spatial coverage |
| ⚡ AI Detection | YOLO real-time object detection for precise obstacle identification |
| 📱 Mobile App | Android companion with live audio alerts & navigation feedback |
| 🔧 Embedded Core | C++ on Arduino IDE for ultra-low-latency performance |
| 🏅 Achievement | Selected for VIT Amaravati Internal Expo |
AI-powered platform connecting farmers and buyers through intelligent crop analysis
| 🔬 ML Pipeline | Crop yield prediction — TensorFlow + Pandas |
| 👁️ Computer Vision | OpenCV real-time crop disease detection |
| 🌐 Backend | Django/Flask marketplace connecting farmers with buyers |
Zero-effort expense tracking by securely parsing your bank transaction SMS
| 💬 Core | Securely parses bank SMS for automatic expense categorization |
| 📈 Dashboard | Real-time financial insights with visual breakdowns |
| 🔐 Security | Firebase auth + encrypted cloud backup |
Emotion intelligence meets medical data — sentiment analysis + real-time report parsing
| 💬 NLP Engine | TensorFlow sentiment analysis for real-time emotional state detection |
| 📄 OCR Pipeline | Pytesseract extracts & structures data from medical reports |
| 🔗 Intelligence | Cross-references health info with curated medical datasets |
Your AI-powered symptom tracker, medication guide, and health companion
| 🔍 Symptom Match | Fuzzy logic symptom matching via FuzzyWuzzy |
| 💊 Health Tracking | Symptoms, medications & personalised diet recommendations |
| 🖥️ Interface | Clean Tkinter GUI with CSV dataset integration |
| 🏅 | 🎓 | 🤖 | 🔗 | 🔧 | 📱 |
|---|---|---|---|---|---|
| VIT Internal Expo | CGPA 8.35 | AI & ML | LLM Infra | Embedded IoT | Mobile Dev |
| Smart Vision Aid Selected | B.Tech CS · 2028 | RAG · Agents · YOLO | LangGraph · Qdrant · RL | ESP32 · Arduino | Flutter · Kotlin · Firebase |
╔══════════════════════════════════════════════════════════════════╗
║ 🧭 On My Workbench ║
╠══════════════════════════════════════════════════════════════════╣
║ 🧠 Cyborg AI — Local RAG System · Flutter UI · LangGraph ║
║ 📈 NewsAlphaAI — LangGraph Stock Prediction ✅ Shipped ║
║ 🤖 Autonomous AI Agent Systems with Tool Use ║
║ 📚 Retrieval-Augmented Generation (RAG) Pipelines ║
║ ⚡ Scalable Backends · FastAPI / Django / Node ║
║ 🔗 LLM Orchestration · LangGraph / LangChain ║
╚══════════════════════════════════════════════════════════════════╝
