Srinath S

AI/ML engineer and 2026 B.Tech Artificial Intelligence & Data Science graduate from Anna University, focused on building applied AI systems with RAG, NLP, vector search, FastAPI, PyTorch, and production-oriented ML workflows.

I build projects that connect model behavior with real software systems: repository intelligence, semantic ticket routing, API-based model inference, and data pipelines. I am comfortable reading documentation, debugging independently, writing clean code, and improving when I get stuck.

Current focus

• Generative AI and Retrieval-Augmented Generation (RAG)
• NLP with embeddings, semantic search, and vector databases
• PyTorch model training and model evaluation
• FastAPI services for AI/ML inference
• Cloud deployment basics with AWS EC2 and Docker
• Clean documentation for recruiter and engineering review

Featured projects

1. Semantic AI Ticket Intelligence Platform

Production expansion of research presented at IEEE ICIRCA 2026.

• Built around SBERT embeddings, FAISS vector search, and confidence-based semantic routing.
• Expanding into a production-oriented system with PyTorch multi-task learning, FastAPI inference, Weights & Biases tracking, and a balanced 108K+ ticket dataset.
• Predicts Category, Team, Priority, and ETA from support ticket text.

Repository: https://github.com/srinath2934/An-End-to-End-Semantic-AI-System-for-Automated-Support-Ticket-Handling

2. RepoChat - GitHub Repository RAG Assistant

A RAG system for asking natural-language questions over GitHub repositories.

• Implements repository ingestion, source chunking, embedding generation, vector retrieval, and LLM-grounded answers.
• Uses Python, LangChain, Hugging Face / SBERT embeddings, vector search, Groq Llama, Streamlit, and Dockerized infrastructure.
• Designed to help developers understand large codebases faster with source-grounded responses.

Repository: https://github.com/srinath2934/RepoChat

3. YOLOv9 Industrial Object Detection

Production-oriented computer vision inference project.

• Built a PyTorch object detection pipeline using YOLOv9.
• Developed FastAPI inference APIs and deployed with Docker on AWS EC2.
• Reduced inference latency from 3-5 seconds to 110ms through model-serving optimization.

4. Credit Risk Classification

Machine learning model for credit risk prediction.

• Built with Python, CatBoost, Scikit-learn, feature engineering, and SHAP explainability.
• Achieved 90% accuracy and 0.74 ROC-AUC.

Technical skills

AI/ML: Python, PyTorch, Scikit-learn, CatBoost, model evaluation, feature engineering, SHAP
Generative AI: LangChain, RAG, LLM APIs, prompt engineering, Hugging Face Transformers
NLP and retrieval: SBERT, embeddings, semantic search, FAISS, vector search
Backend and deployment: FastAPI, REST APIs, Docker, AWS EC2, Linux, Git, GitHub
Data: SQL, Pandas, NumPy, data cleaning, statistical analysis, Power BI, Excel

Certifications

• LangChain Certification - LLM application and RAG workflow development
• Introduction to Model Context Protocol (MCP) - Anthropic
• IBM Python for Data Science, AI & Development
• IBM Python Project for Data Science
• SQL (Intermediate) - HackerRank

Education

B.Tech - Artificial Intelligence & Data Science
JCT College of Engineering & Technology, Anna University
2022 - 2026 | CGPA: 8.0 / 10

Connect

LinkedIn: https://www.linkedin.com/in/srinath29
Email: srinath2934@gmail.com