AI/ML Engineer Vision, OCR, RAG, Generative AI

Location:

Posted:

June 13, 2026

TEJA KUMAR G S

+91-638******* ************@*****.*** LinkedIn Github Portfolio

CAREER OBJECTIVE

AI Developer with hands-on internship experience in industrial computer vision, YOLO-based object

detection, RAG pipelines, and OCR systems — seeking an entry-level AI/ML role to contribute to real-world

AI systems while deepening expertise in deep learning and generative AI.

TECHNICAL SKILLS

• Programming: Python

• Computer Vision: Ultralytics YOLO, OpenCV, industrial video analytics

• LLMs & GenAI: LangChain, Ollama, Mistral, LLaMA, RAG, prompt engineering, MCP

• ML & NLP: Hugging Face Transformers, PyTorch, PaddleOCR, Tesseract

• Deployment & Databases: Flask, FastAPI, Docker, ChromaDB, MySQL

• Tools & Annotation: Git, GitHub, CVAT, data augmentation, YOLO dataset formatting

PROFESSIONAL EXPERIENCE

AI Intern at Defect Scanner JUL 2025 – JAN 2026

• Designed and deployed a YOLO-based defect detection model on (mAP50: 0.92) 17K+ images, optimized

to 13+ FPS with an end-to-end OpenCV pipeline handling real-world noise, blur, and lighting variation.

• Engineered spatial-temporal validation logic (centroid tracking, ROI checks) and an automated alert

system (tower lamp + buzzer), achieving ~95% process compliance (~60 units/shift) and reducing

manual inspection effort by ~40%.

• Deployed model as a real-time Flask REST API with Docker-ready modular architecture, integrating

augmentation strategies and quality validation across a 90K+ annotation training pipeline built in CVAT.

DL Engineer Intern at FDAI DEC 2024 – JUN 2025

• Prototyped RAG pipelines using locally deployed LLMs (Ollama, Mistral) with LangChain combining

OCR extraction, embeddings, and prompt engineering for document summarization and Q&A.

• Integrated Hugging Face Transformers, PaddleOCR, and Tesseract for NLP and multimodal document

understanding workflows including text extraction and entity recognition.

• Deployed AI models in secure air-gapped on-premises environments and built FastAPI-based demo

applications exposing model functionality as RESTful endpoints for stakeholder presentations.

PROJECTS

RAG-Powered Document Intelligence System LangChain · ChromaDB · Ollama · Mistral · LLaMA

• Built an End-to-end RAG pipeline for document ingestion, semantic chunking, embedding generation,

vector storage and AI-powered Knowledge retrieval

Image based PDF Digitization pipeline PaddleOCR · Tesseract · OpenCV

• Built an image-based PDF OCR pipeline using PaddleOCR and Tesseract to extract text, tables, figures,

and links — enabling scalable document digitization.

EDUCATION

MSc. IT- University of Madras, Guindy, Chennai Aug 2023 – May 2025

BSc. CS – Government Arts College, Nandanam, Chennai SEP 2020 – JUN 2023