Skip to main content

Swaraj Chattaraj
AI Systems Engineer

I build production AI systems — agentic pipelines, retrieval-augmented reasoning, and the infrastructure that runs them at scale.

View Projects Resume (PDF) Contact GitHub
Progression

Engineering
Timeline

Python & Algorithms

Data structures, OOP, and algorithmic foundations in Python, C, C++, and Java.

Phase 01
Phase 02

Machine & Deep Learning

TensorFlow, PyTorch, Scikit-learn, XGBoost. Training classical models and neural networks.

LLMs & RAG

HuggingFace, ChromaDB, FAISS. Building semantic and hybrid retrieval systems.

Phase 03
Phase 04

Agentic AI

LangGraph, Multi-Agent Pipelines. Developing reasoning loops, self-correction, and Critic evaluators.

Production & Cloud

FastAPI, React, Azure, Docker, Vercel, Render. Shipping robust full-stack AI architectures.

Current
Flagship System

ARIA

Autonomous Research & Intelligence Assistant. A 5-stage agentic RAG pipeline built with LangGraph, reducing retrieval latency by 63% via concurrent hybrid search across Wikipedia, OpenAlex, arXiv, and DuckDuckGo.

GitHub Repo Explore ARIA
ARIA Dashboard // React SPA // JWT Auth
Session Log
> Executing query...
> 4 sources active
> Latency: 4s (~63% reduction)
LangGraph Orchestrator Pipeline
Plan
Retrieve
Synthesize
Audit
Publish
63% Latency Reduction

Concurrent hybrid retrieval (`asyncio`, `aiohttp`) cutting retrieval time from ~11s to ~4s across 4 knowledge sources.

11/11 Unit Tests Passing

Ragas-based eval suite verifying faithfulness, answer relevancy, and context precision. 50% leakage tests passed.

Full Cross-Platform Stack

Azure VM (Ubuntu + Caddy). React/Tailwind SPA, PyWebView Desktop app, Gradle/TWA Android APK. PDF exports via PyMuPDF.

Pipeline Infrastructure

RAGPRO

A high-performance modular RAG pipeline focused on sub-100ms vector search indexing, local offline embeddings, and seamless LLM integration.

View Repository
1
Upload PDF / DOCX
CSV / MD
2
Chunk/Embed HuggingFace
all-MiniLM-L6-v2
3
FAISS Store Local Vector
Persistence
4
Retriever FastAPI REST
Sub-100ms
5
LLM OpenRouter
100+ Models
6
Answer Grounded &
Offline-Capable

Technical Skills

Python
LangGraph
Azure
React
FastAPI
Docker
TensorFlow
PyTorch
FAISS
ChromaDB
MongoDB
SQL

Languages

  • Python, C, C++
  • Java, SQL

Agentic / Orchestration

  • LangGraph, LangChain
  • OpenAI API, Azure OpenAI
  • OpenRouter, HuggingFace

ML / NLP

  • TensorFlow, PyTorch
  • Scikit-learn, XGBoost
  • BERT, NLTK, spaCy

Retrieval & Eval

  • ChromaDB, FAISS
  • Semantic / Hybrid Search
  • Ragas (Faithfulness)

Backend

  • FastAPI, Uvicorn
  • JWT Auth
  • asyncio, aiohttp

Frontend

  • React, Vite, Tailwind
  • Streamlit, PyWebView
  • Gradle / TWA

Cloud & MLOps

  • Azure, AWS, GCP
  • Render, Vercel
  • Docker, CI/CD, Caddy

Data & Tools

  • Pandas, NumPy
  • Matplotlib, Git
  • PyMuPDF, ReportLab
Future Focus

Research
Interests

Beyond building systems, I am focused on evaluating, orchestrating, and securing agentic models. These are my core areas of interest for academic and professional exploration.

Agentic AI +
Designing self-operating pipelines where models plan, execute, and evaluate their own tool usage autonomously.
LLM Evaluation +
Moving beyond vibes to strict, deterministic metrics for faithfulness, context precision, and hallucination rates (e.g. using Ragas).
Multi-Agent Systems +
Orchestrating distinct specialized agents (Planner, Researcher, Critic) in LangGraph to achieve complex goal resolution.
AI Infrastructure +
Scaling inference, optimizing vector store indexing (FAISS/ChromaDB), and deploying async real-time REST layers.
Reasoning Models +
Enhancing deterministic logic inside probabilistic models using Chain-of-Thought and programmatic guardrails.
Let's collaborate

Let's build intelligent software together.

Open to internships, research collaborations, and engineering roles in Agentic AI and Cloud infrastructure. Reach out to discuss technical architectures or team opportunities.

swarajchattaraj17402@gmail.com Resume
GitHub LinkedIn