VaibhavLabs.ipynb

[ ]:

from vaibhav import MLEngineer
 
vaibhav = MLEngineer(
    name="Vaibhav R. Dandala",
    university="NC State University",
    degree="B.S. EE + CPE (Dual)",
    gpa=3.80,
    focus=["Agentic LLM/RAG", "Building End-To-End Tools", "Computer Vision"],
)
vaibhav.display()

Vaibhav R. Dandala

ML Engineer · AI Researcher · Research Engineer

Agentic LLM/RAG Pipelines · Building End-To-End Tools · Computer Vision · Production AI Systems

EE + CPE (Dual Degree)US CitizenNCSU '28

GitHub LinkedIn Resume Email

[ ]:

vaibhav.experience.timeline(verbose=True)

Achieved 93% accuracy on vertebrae segmentation and lesion detection by co-developing a CNN pipeline combining MedSAM3 for segmentation and YOLOv8 for detection, trained on 65,000+ CT images.
Reduced radiologist data prep time by 35% across active clinical trials by building segmentation and preprocessing pipelines, directly accelerating oncology research throughput.

[ ]:

vaibhav.publications.all()

LUMIN

ICMLA 2025

Agentic LLM Multimodal NASA PDS Query System

Co-authored with Dr. Emily Dunkel · NASA Jet Propulsion Laboratory

Hybrid RAG workflow over MSL (Mars Science Laboratory) archives using BM25 + dense embeddings. Enables planetary scientists to query decades of Mars rover data through a natural language interface. Achieved 80%+ PDS3 filter prediction accuracy across 10,000+ indexed mission records.

RAGLLMNASAMultimodalBM25PineconeDB

View Paper

[ ]:

vaibhav.projects.head(3)

AI-native CRM and marketing automation platform for realtors. Agentic inbox, campaign sequences, and LinkedIn/X marketing. Built Key Moments — a pg_cron-scheduled milestone engine that invokes Claude with pgvector RAG to generate personalized drafts, with approval-gated agentic writes and immutable per-event audit traces.

100+

Active realtors

Doorify

MLS partnership

Approval-gated + audit-traced

Agentic writes

What was technically hard

Building approval-gated agentic writes with immutable audit traces at the per-event level — ensuring every Claude-generated draft was traceable, reversible, and safe for real client communication.

Next.jsReactSupabasePostgreSQLpgvectorGeminiClaudeGmail APILinkedIn APIX APIChrome MV3

[ ]:

%%bash
echo "=== Contact Vaibhav ==="
cat contact.txt

$ echo "=== Contact Vaibhav ==="

=== Contact Vaibhav ===

$ cat contact.txt

Email : vrdandal@ncsu.edu

GitHub : github.com/VaibhavD2006

LinkedIn : linkedin.com/in/vaibhav-dandala

Phone : (984) 244-8318

$ mail vaibhav

5 cells in Portfolio

learning.ipynb

# markdown

Papers & Techniques I'm Exploring

[ ]:

papers = {
    "MedSAM3": "Foundation model for medical image segmentation (Meta AI, 2024)",
    "Image Detection": "Fine-tuning detection models for Oncology + US Defense",
    "Hybrid RAG": "BM25 sparse + dense embeddings with re-ranking",
}
papers

"MedSAM3": "Foundation model for medical image segmentation (Meta AI, 2024)"

"Image Detection": "Fine-tuning detection models for Oncology + US Defense"

"Hybrid RAG": "BM25 sparse + dense embeddings with re-ranking"

# markdown

Currently Building

[ ]:

# Quant Finance
quant = {
    "VIX Prediction Bot": "$500 P&L realized (not paper trade)",
    "Other strategies": [
        "Post Earnings Drift Announcement via SEC filing web scrapers",
        "Options flow + RSI mean reversion across sectors",
        "Monte Carlo simulations for risk analysis",
    ],
    "stack": {
        "runtime":    "Python 3.x",
        "data":       "DuckDB + Polars + PyArrow + yfinance + EDGAR",
        "signals_ml": "scikit-learn + hmmlearn + statsmodels + cvxpy",
        "serving":    "FastAPI + uvicorn",
        "tracking":   "MLflow",
    },
}
quant

"VIX Prediction Bot": "$500 P&L realized (not paper trade)"

"Other strategies": [

"Post Earnings Drift Announcement via SEC filing web scrapers",

"Options flow + RSI mean reversion across sectors",

"Monte Carlo simulations for risk analysis",

]

"stack": {

"runtime": "Python 3.x"

"data": "DuckDB + Polars + PyArrow + yfinance + EDGAR"

"signals_ml": "scikit-learn + hmmlearn + statsmodels + cvxpy"

"serving": "FastAPI + uvicorn"

"tracking": "MLflow"

}

[ ]:

# CaseForge AI
caseforge = {
    "description": "AI-powered consulting case interview prep SaaS",
    "target": "NC State consulting clubs — McKinsey, BCG, Bain, T2 firms",
    "features": [
        "AI Interviewer — adaptive mock interviews with pressure testing",
        "7-Dimension Scorecard — structure, hypothesis, numerics, synthesis, comms",
        "Readiness Dashboard — Bronze → Silver → Gold → Platinum tier progression",
        "Drill Library — 100+ cases across retail, healthcare, M&A, operations",
        "Session History & Analytics — charts and dimension trends over time",
    ],
    "stack": {
        "framework": "Next.js 16 + TypeScript + React 19",
        "styling":   "Tailwind CSS v4 + Shadcn UI + Framer Motion",
        "ai":        "Google Gemini + Ollama (local inference)",
        "db":        "PostgreSQL + Drizzle ORM",
        "auth":      "Auth.js v5 — Google OAuth",
        "jobs":      "Inngest — 7-stage async eval pipeline",
    },
}
caseforge

"description": "AI-powered consulting case interview prep SaaS"

"target": "NC State consulting clubs — McKinsey, BCG, Bain, T2 firms"

"features": [

"AI Interviewer — adaptive mock interviews with pressure testing",

"7-Dimension Scorecard — structure, hypothesis, numerics, synthesis, comms",

"Readiness Dashboard — Bronze → Silver → Gold → Platinum tier progression",

"Drill Library — 100+ cases across retail, healthcare, M&A, operations",

"Session History & Analytics — charts and dimension trends over time",

]

"stack": {

"framework": "Next.js 16 + TypeScript + React 19"

"styling": "Tailwind CSS v4 + Shadcn UI + Framer Motion"

"ai": "Google Gemini + Ollama (local inference)"

"db": "PostgreSQL + Drizzle ORM"

"auth": "Auth.js v5 — Google OAuth"

"jobs": "Inngest — 7-stage async eval pipeline"

}

5 cells in Learning Journal

Run Selected Cell	Shift+Enter
Command Palette	Ctrl+K
Close dialogs	Esc