Shourya Roy

B.Tech Computer Science and Engineering graduate (Minor in Machine Learning & Data Analytics) with industry experience working as a Generative AI and Artificial Intelligence Engineer across internships and professional roles. Proven ability to design and deploy LLM-powered systems, RAG architectures, and agentic AI platforms using Python, OpenAI APIs, FastAPI, and modern AI frameworks. Experienced in Deep Learning, NLP, Computer Vision, and Data Analytics, with hands-on experience building full-stack AI applications.

Generative AI Engineer at Growhut (2025-10 – 2026-04)

• Designed and deployed LLM-powered chatbots and RAG-based systems using OpenAI and Gemini APIs, incorporating advanced prompting strategies, structured outputs and tool-augmented workflows via Composio.
• Built retrieval and data-processing services using hybrid search, embeddings, reranking, and vector databases, improving response relevance and system performance.
• Integrated memory frameworks Vector/Graph Databases including SuperMemory, Mem0, Qdrant, Pinecone, and Neo4j to enable contextual and persistent AI interactions.
• Developed a multilingual meeting transcription system featuring automatic speech recognition (fasterwhisper large-v3), regional language translation (IndicTrans2), and LLM-based transcript cleanup — with optional speaker diarization for multi-speaker scenarios.
• Implemented OCR-based text extraction workflows using Tesseract for processing scanned documents and images.
• Developed scalable backend services using FastAPI and optimized Python workflows for automation and performance.
• Developed and maintained REST APIs and backend services handling document processing, transcription, and workflow automation.
• Collaborated with cross-functional teams in an Agile development environment, participating in sprint planning, regular stand-ups, and iterative feature delivery.
• Leveraged AI developer tools (Cursor, Claude Code) to accelerate development, debugging, and code generation workflows.
• Managed end-to-end project delivery, including client interaction.
• Utilized Docker and Git for deployment and version control.

Intern Software Developer at Qapp.ai (2025-07 – 2025-10)

• Developed a full-stack Machine Learning web application using Python, Flask, and Vue.js to train and visualize Decision Tree models on user-provided datasets.
• Built a dynamic and reactive front-end using Vue.js, JavaScript, Bootstrap, HTML, and CSS, enabling real-time UI updates and interactive user workflows.
• Implemented responsive components for data input, model training, and visualization, improving usability and user interaction.
• Developed the application using an iterative development approach, incrementally delivering components, incorporating feedback, and continuously improving system functionality.

AI Intern at Atos (2025-04 – 2025-07)

• Designed and implemented an Agentic AI platform using Python, Generative AI and AutoGen for autonomous factory monitoring.
• Enabled predictive maintenance and failure prevention through multi-agent collaboration systems.

Computer Vision Intern at C-DOT (2025-03 – 2025-04)

• Developed a real-time driver drowsiness detection system using Python and YOLOv8.
• Built and optimized real-time inference pipelines using Python and OpenCV for low-latency processing.
• Integrated video processing pipelines for handling and preprocessing real-time video streams.
• Developed and tested the application on Linux systems, leveraging GPU acceleration for efficient deep learning inference and real-time video analytics.
• Applied Computer Vision, Machine Learning, and Data Analytics methodologies.

AI/NLP Intern at Eviden (2024-07 – 2024-08)

• Built an end-to-end multi-level document classification system using Machine Learning, NLP and Data Analytics techniques to categorize medical documents.
• Generated page-level summaries and titles using GenAI frameworks.
• Used Python libraries like Spacy, nltk, Pandas and HuggingFace
• Achieved 100% accuracy and F1-score through optimized NLP workflows using LangChain.

ML Intern at C-DOT (2023-07 – 2023-08)

• Developed predictive crop pattern models in Python, comparing multiple algorithms: Logistic Regression (96.14%), Neural Networks (96.36%), and Decision Trees (99.31%) to determine optimal crop predictions.
• Used Python libraries like Scikit-learn and Tensorflow

B.Tech in Computer Science and Engineering with Minor Specialization in Machine Learning and Data Analytics – Maharaja Agrasen Institute of Technology (IP University, Delhi) (2021 – 2025)

Class XII in CBSE Board – St. Columba's School, New Delhi (2021 – 2021)

Class X in CBSE Board – St. Columba's School, New Delhi (2019 – 2019)