██╗ ██╗ █████╗ ██████╗ ████████╗██╗ ██╗██╗██╗ ██╗
██║ ██╔╝██╔══██╗██╔══██╗╚══██╔══╝██║ ██║██║██║ ██╔╝
█████╔╝ ███████║██████╔╝ ██║ ███████║██║█████╔╝
██╔═██╗ ██╔══██║██╔══██╗ ██║ ██╔══██║██║██╔═██╗
██║ ██╗██║ ██║██║ ██║ ██║ ██║ ██║██║██║ ██╗
╚═╝ ╚═╝╚═╝ ╚═╝╚═╝ ╚═╝ ╚═╝ ╚═╝ ╚═╝╚═╝╚═╝ ╚═╝
██╗ █████╗ ██████╗ ██╗ ██╗ █████╗ ██╗ ██╗
██║██╔══██╗██╔══██╗██║ ██║██╔══██╗██║ ██║
██║███████║██║ ██║███████║███████║██║ ██║
██ ██║██╔══██║██║ ██║██╔══██║██╔══██║╚██╗ ██╔╝
╚█████╔╝██║ ██║██████╔╝██║ ██║██║ ██║ ╚████╔╝
╚════╝ ╚═╝ ╚═╝╚═════╝ ╚═╝ ╚═╝╚═╝ ╚═╝ ╚═══╝
class KarthikJadhav:
def __init__(self):
self.university = "IIIT Vadodara, Gujarat"
self.degree = "B.Tech Computer Science"
self.focus = ["RAG Systems", "LLMs", "Information Retrieval"]
self.currently = "Building production-grade AI retrieval pipelines"
self.fun_fact = "I improve recall@k, not just vibes 📈"
def say_hi(self):
print("Hey! I build AI systems that actually retrieve the right stuff.")AI / ML Stack
Retrieval & Vector Systems
Full Stack & Tools
The thing I'm most proud of.
A multi-stage Retrieval-Augmented Generation system that goes way beyond naive vector search.
Query → Expansion (LLM) → Dense Search (ChromaDB) + Sparse Search (BM25)
→ Reciprocal Rank Fusion → Cross-Encoder Re-ranking → Grounded Answer
| Metric | Improvement |
|---|---|
| Retrieval Relevance | ↑ 40–60% |
| Recall (via RRF) | ↑ 2.3× |
| Irrelevant Results | ↓ 35% |
| Documents Supported | 1000+ |
Tech: Python Gemini ChromaDB BM25 Sentence-Transformers Arize Phoenix Streamlit
AI that helps governments actually listen to citizens.
A full-stack platform that turns raw citizen complaints + policy docs into actionable governance insights.
- 🗺️ Geospatial heatmaps to pinpoint high-impact issue clusters
- 🤖 Gemini AI pipeline for complaint classification & sentiment analysis
- 📄 Document intelligence for summarizing long government reports
- 🎙️ LLM-generated public communications from live data
Tech: React TypeScript Node.js PostgreSQL Gemini AI
--
| Certification | Issuer | Valid Until |
|---|---|---|
| ☁️ Oracle Cloud Infrastructure 2025 AI Foundations Associate | Oracle | Oct 2027 |
Information Retrieval ████████████████████ Vector Search
Query Expansion ████████████████░░░░ Re-ranking
Embeddings ███████████████░░░░░ Evaluation (Precision@k)
Deep Learning ██████████████░░░░░░ LLM Systems
RAG Architecture █████████████████░░░ Observability
"Retrieval is not search. It's the art of finding what the user actually meant."