UAV

AI ENGINEER / GENERATIVE AI

Udhaya Adhithya Varman

AI Engineer and CSE (AI) graduate building end-to-end GenAI systems: hybrid RAG with hallucination checks, multi-agent research pipelines, and real-time edge-deployed ML.

Explore Projects Get in Touch

01 / BACKGROUND

About Me

I'm a Computer Science and Engineering (Artificial Intelligence) graduate with a focus on building practical AI applications. My work spans retrieval-augmented generation, multi-agent research workflows, and real-time computer vision deployed at the edge.

I enjoy turning machine-learning ideas into end-to-end systems—from retrieving relevant knowledge and checking claims to building usable interfaces and optimizing inference speed.

02 / PROFESSIONAL EXPERIENCE

Experience

AI Engineer Intern

23 Jul 2026 – Present

Unizen Technologies · Bengaluru, India

  • Contributing to live projects across artificial intelligence, machine learning, and embedded systems.
  • Applying technical skills to real-world engineering tasks while collaborating with experienced professionals.

03 / SELECTED WORK

Projects

Featured AI systems spanning hybrid dense-sparse retrieval, multi-agent collaborative workflows, and edge neural architectures.

Domain-Specific AI Research Assistant with Knowledge Graph

A research assistant that ingests academic PDFs and answers questions using hybrid RAG. Combines ChromaDB dense retrieval with BM25 sparse retrieval through Reciprocal Rank Fusion, checks generated claims using an NLI cross-encoder, and builds an interactive multi-paper knowledge graph across models, datasets, and metrics.

  • Ollama
  • Hybrid RAG
  • ChromaDB
  • BM25
  • Streamlit
View Project on GitHub

Multi-Agent AI Research Assistant

Five specialized agents—Manager, Searcher, Summarizer, Critic, and Writer—automate academic research using Semantic Scholar. The pipeline ranks papers by relevance, citation impact, and recency, then generates structured literature reviews with gap analysis and citation validation.

  • LLaMA 3.3-70B
  • Groq
  • ChromaDB
  • Semantic Scholar
  • FastAPI
  • Streamlit
View Project on GitHub

SignBridge — Real-Time ISL Translator with Edge Deployment

Indian Sign Language recognition system for all 26 ISL alphabet signs. Uses a sliding 30-frame CNN–LSTM model with multi-head attention and MediaPipe hand landmarks. Reports 99.74% validation accuracy, 22-minute CPU training, and 353 FPS with ONNX FP32 deployment.

  • MediaPipe
  • CNN–LSTM
  • ONNX
  • OpenCV
  • pyttsx3
View Project on GitHub

04 / EXPERTISE & BACKGROUND

Skills

Languages & Tools
Python, SQL, Streamlit
AI & Machine Learning
Artificial Intelligence, Machine Learning, NLP, Deep Learning, LangGraph, Data Analysis
Generative AI Projects
Hybrid RAG, BM25, ChromaDB, NLI-based hallucination checking, multi-agent research systems
Computer Vision
MediaPipe, CNN–LSTM, ONNX, OpenCV
Cloud & MLOps
MLflow, Docker, GitHub Actions

05 / EDUCATION & CERTIFICATIONS

Education

Academic Background

Aug 2022 – Jul 2026

Sathyabama Institute of Science and Technology of Engineering

Bachelor of Computer Science Engineering (CSE) – AI

CGPA: 8.0

Certifications

  • Certification in Artificial Intelligence
  • Certification in Deep Learning Fundamental
  • Certification in Machine Learning
  • Certification in Prompt Engineering for Everyone
  • Certification in Python for Database

06 / CONTACT

Let's build something

Interested in collaborating or discussing an AI engineering opportunity? Reach out by email or connect with me online.

udhayaadhithyavarman@gmail.com

+91 9483215500 India

GitHub LinkedIn Resume (PDF)