█▀▄▀█ █▀█ █▀▄ █▀▀ █▀▄▀█ █▀▄ █ ▄▀█ █ █ █ █▀█ █ █▄ █
█ ▀ █ █▄█ █▄▀ ██▄ █ ▀ █ █▄▀ █ █▀█ █▄▄ █▄█ █▀▀ █ █ ▀█
▄ NOW LOADING ▄ ▄ ▄ PLEASE STAND BY ▄
class AIEngineer:
def __init__(self):
self.name = "Satyam"
self.role = "AI/ML Engineer"
self.status = "Building intelligent systems that actually make sense"
self.location = "United Institute of Technology"
self.languages = ["Python", "JavaScript", "TypeScript"]
self.current_obsession = "Multi-agent AI orchestration"
def what_i_do(self):
return [
"🤖 Build LLM-powered systems that understand context",
"🔗 Design RAG pipelines that retrieve the right info",
"🎯 Orchestrate AI agents that collaborate like a team",
"⚡ Deploy ML models that work in production (not just notebooks)"
]
def philosophy(self):
return "AI is only as smart as the context it gets — that's why I obsess over RAG"
def vibe_check(self):
return "Coffee + Code + Curiosity = 🔥"Currently: B.Tech student specializing in AI & ML | Building systems that learn, not just memorize
Vibe: Equal parts researcher, builder, and someone who reads AI papers at 2 AM because they're genuinely interesting
📼 SIDE A: AI & LLM TOOLKIT
╔══════════════════════════════════════╗
║ 🎙️ LLM & AI FRAMEWORKS v3.0 ║
╠══════════════════════════════════════╣
║ ║
║ 🧠 OpenAI GPT Models ║
║ 🔗 LangChain (chain of thought) ║
║ 🤝 CrewAI (multi-agent magic) ║
║ 🤗 HuggingFace (transformers FTW) ║
║ 💬 Prompt Engineering (the art) ║
║ ║
╚══════════════════════════════════════╝
📼 SIDE B: CORE ENGINEERING
╔══════════════════════════════════════╗
║ ⚙️ BACKEND & DATA STACK v2.0 ║
╠══════════════════════════════════════╣
║ ║
║ 🐍 Python (primary weapon) ║
║ ⚡ FastAPI (async everything) ║
║ 🔍 FAISS (vector search FTW) ║
║ 📊 Pandas + NumPy (data wrangling) ║
║ 🐳 Docker (containers everywhere) ║
║ ☁️ AWS (cloud deployments) ║
║ ║
╚══════════════════════════════════════╝
|
AI agents working together like a startup team Tech: CrewAI, LangChain, OpenAI |
Making LLMs actually know what they're talking about Tech: FAISS, LangChain, Python |
|
Talk to your documents (literally) What it does: Why it's cool: |
AST parsing meets LLM reasoning The combo: Result: |
|
Know who's leaving before they do ML pipeline that flags at-risk users using behavioral signals — so you can intervene before it's too late Tech: XGBoost, SHAP, FastAPI |
Can you tell if a human wrote this? Hybrid classifier (perplexity + stylometric features) that catches AI-generated content across GPT-4, Claude, and Gemini outputs Tech: Transformers, FastAPI, Streamlit |
const currentStatus = {
focus: "Multi-agent systems + advanced RAG optimization",
learning: ["LoRA fine-tuning", "LLM evaluation frameworks", "agentic patterns"],
building: "Production AI systems that scale",
reading: "Research papers on retrieval augmentation",
mood: "Caffeinated and curious",
last_commit: "Probably refactoring embeddings pipeline",
music: "lofi beats + AI agent orchestration = flow state"
};Open to collab on:
- 🤖 Multi-agent systems
- 📚 Document intelligence
- 🔍 Advanced RAG architectures
- 🚀 Production ML deployments
"The best AI systems are the ones that know when to say 'I don't know'"
— Me, after fixing hallucination issues
Dev Musings:
- 📟 RAG is basically giving your AI a library card
- 📟 Multi-agent systems are like herding cats, but the cats are GPT models
- 📟 Embeddings are just fancy word math (but the math is beautiful)
- 📟 "It works on my machine" hits different when deploying ML models
- 📟 Vector databases are underrated and deserve more love
Current vibe: Building systems that understand context, one vector at a time
📼 AI Developer Intern @ Asvix (Jan 2026 - Apr 2026)
ACHIEVEMENTS UNLOCKED:
├─ 🧠 Built DigiLab RAG pipeline — 500+ daily queries, 99.2% uptime
├─ 📉 Cut hallucination rate 39% (18% → 11%)
├─ 📊 RAGAS evaluation harness for ongoing quality tracking
└─ 🔗 LangChain + FAISS + Neo4j hybrid RAG
Translation: Built a production RAG system that actually stayed up and kept getting better
📼 AI Chatbot Development Intern @ Cloudily Scripts (Jun 2025 - Jul 2025)
ACHIEVEMENTS UNLOCKED:
├─ 🎯 Built chatbots that reduced tickets by 35%
├─ 📈 Improved accuracy from 72% → 91%
├─ ⚡ Got latency under 2 seconds
├─ 🐳 Dockerized everything
└─ 🔄 Implemented end-to-end RAG workflows
Translation: Made AI chatbots that actually help instead of frustrate
☁️ Cloud Engineering Intern @ IPtechhub (May 2024 - Jul 2024)
ACHIEVEMENTS UNLOCKED:
├─ ☁️ Deployed ML services on AWS
├─ ⚙️ Built CI/CD pipelines
├─ 📊 Handled 3x traffic spikes without breaking
└─ 🤖 Automated everything that could be automated
Translation: Learned that deployment is where the real challenges hide
┌──────────────────────────────────────┐
│ MODEM STATUS: ONLINE │
│ PROTOCOL: HTTP/2.0 │
│ SIGNAL: ▰▰▰▰▰▰▰▰▱▱ 85% │
│ LATENCY: 42ms (nice) │
└──────────────────────────────────────┘
| Metric | Status |
|---|---|
| Coffee Consumed | ∞ |
| AI Papers Read | Too many to count |
| Production Bugs Fixed | More than I'd like to admit |
| Embeddings Generated | Millions |
| RAG Pipelines Built | Still building |
| "It works on my machine" | Occasionally |