Aswini Rajulapudi
Andhra Pradesh
Summary
AI/ML Engineer with hands-on experience building multi-agent LLM systems and RAG pipelines in production. Currently designing finance-domain multi-agent AI chatbots at Big Air Lab using custom Python SDK architectures.
Overview
1
1
years of professional experience
Work History
AI/ML Intern
Big Air Lab
Bengaluru
06.2025 - Current
- Contributing to the development of a finance-based AI chatbot product, built using agent-based LLM architecture.
- Initially worked with LangGraph, LangChain to design conversational flows, prompts, and structured outputs using Pydantic schemas.
- Identified limitations in LangGraph regarding LLM internal reasoning visibility and token usage tracking, leading to a major architectural shift. So moved to New Architecture design with custom Python SDK-based agent implementation.
- Built and integrated:
- - Gatekeeper Agent for user query validation and safety checks
- - Orchestrator Agent to manage and route tasks across internal agents
- - Synthesizer Agent to aggregate multi-agent outputs into a coherent final response
- Moved to custom SDK architecture, reducing token tracking overhead and improving response transparency.
Associate AI/ML Intern
Rava.ai
Hyderabad
12.2024 - 04.2025
- Built RAG pipelines using FAISS for PDF and URL-based document retrieval.
- Designed and executed 30+ pytest test suites for LLM workflows and backend modules.
- Developed a lightweight HTML/JavaScript interface to test and validate LLM responses.
- Remote, T-Hub, Hyderabad
Summer Internship
NIT Warangal
India
05.2024 - 06.2024
- Enhanced Phishing Detection and Probabilistic Risk Analysis for URLs.
- A 'PHISHUSIL Dataset', which have more than 2,00,000 URLs had been taken to build a model using various boosting algorithms. Also explored a few IEEE papers on this project to understand and improve the performance.
- Achieved accuracy of 99.44% using the AdaBoost classifier and 99.43% using XGBoost Algorithm.
- This project ends with the calculation of the risk score of a particular URL based on its attributes which ranges between 0-1, where for phishing url's it will be near to 1 otherwise 0 for legitimate url's.
Education
Bachelor Of Technologies - Computer Science and Engineering
Rajiv Gandhi University of Knowledge Technologies
01.2025
Skills
- Java
- Python
- C
- Supervised Learning Algorithms
- Clustering Algorithms
- Scikit-learn
- PyTorch
- TensorFlow
- CNNs
- NLTK
- Transformers
- Large Language Models
- RAG
- Prompt Engineering
- LangChain
- LangGraph
- Agentic AI
- Model Evaluation
- RAGAS
- DeepEval
- OpenAI Python SDK
- GPT-4
- GPT-5
- Fine-tuning LLMs
- MCP
- FastAPI
- Streamlit
- HTML
- CSS
- SQL
- FAISS Vector DB
- Qdrant Vector DB
- BeautifulSoup
- Git
- GitHub
- Docker
- Pytest
- Linux
- Data Structures
- Algorithms
- Object Oriented Programming
- Team Collaboration
- Leadership
- Problem Solving
- Communication
Websites
Languages
Java
Python
C
Internships Experience
- AI/ML Intern, Big Air Lab, Bengaluru, Karnataka, 2025-06-01, Present, Contributing to the development of a finance-based AI chatbot product, built using agent-based LLM architecture., Initially worked with LangGraph, LangChain to design conversational flows, prompts, and structured outputs using Pydantic schemas., Identified limitations in LangGraph regarding LLM internal reasoning visibility and token usage tracking, leading to a major architectural shift. So moved to New Architecture design with custom Python SDK-based agent implementation., Built and integrated: Gatekeeper Agent for user query validation and safety checks, Orchestrator Agent to manage and route tasks across internal agents, Synthesizer Agent to aggregate multi-agent outputs into a coherent final response., Moved to custom SDK architecture, reducing token tracking overhead and improving response transparency.
- Associate AI/ML Intern, Rava.ai, Hyderabad, Telangana, 2024-12-01, 2025-04-01, Built RAG pipelines using FAISS for PDF and URL-based document retrieval., Designed and executed 30+ pytest test suites for LLM workflows and backend modules., Developed a lightweight HTML/JavaScript interface to test and validate LLM responses.
- Intern, NIT Warangal, Warangal, Telangana, 2024-05-01, 2024-06-01, Enhanced Phishing Detection and Probabilistic Risk Analysis for URLs., Achieved accuracy of 99.44% using the AdaBoost classifier and 99.43% using XGBoost Algorithm.
Extracurricular Activities
Student Coordinator, Training and Placement Cell, RGUKT Srikakulam, Actively organized & participated in on campus placement drives. Event Organizer, Samavedhan Event, Led a team in planning and managing various aspects of the event.
Projects
- Multimodal RAG System (Text + Image), Designed a multimodal RAG pipeline that accepts any file format as input.
- Retrieval Augmented Generation Chatbot, Web URL-based Chatbot built using LangChain and Groq LLMs with FAISS for vector storage.
- Deep Learning Based Effective Reverberation And Noise Suppression For Enhancing The Target Detection In Active Sonar Systems, Explored deep learning techniques to enhance the Signal-to-Reverberation Ratio (SRR).
- Bank Customer Churn Analysis, Developed predictive models to accurately predict customer churn.
Timeline
AI/ML Intern
Big Air Lab
06.2025 - Current
Associate AI/ML Intern
Rava.ai
12.2024 - 04.2025
Summer Internship
NIT Warangal
05.2024 - 06.2024
Bachelor Of Technologies - Computer Science and Engineering
Rajiv Gandhi University of Knowledge Technologies