Professional Experience and Projects
Recommendation System using YOLO-v11 and AR
- Working with Generative AI, Stable-Diffusion for image to image generation, Yolo for Segmentation and Detection
- Working with Label-Studio tool for image annotation
Custom Agentic RAG Chatbot for E-commerce customer support.
Surveillance System using OpenCV and Ultralytics YOLO
Fine-tuned YOLOv8 model with the custom dataset for human detection and used OpenCV for other operations such as counting, saving, optimizing the frame. Used Fast-API for inference.
- Worked with NLP, Computer Vision, LLM, Hugging face,Fine-tuning, transfer learning, Python etc.
- Worked with Fast API, Flask, Celery, .Net core, C#, Microsoft SQL Database, AWS, Lambda, SQS, SNS, MQTT, RabbitMQ, and other latest technologies.
- Co-operation with REST API, GraphQL and SQL Database system.
- Worked with Angular (TypeScript) as frontend technology and had experienced with MAUI for desktop application.
- Worked on various financial and Banking Android applications such as National Life Insurance, Mercantile Bank Ltd.
- Extracted and Implemented data from REST API to Android application using various library such as HTTP Client, Retrofit, Volley Library.
- Used Graph-QL as a new & fast technology.
Syntax Solution Ltd.
Machine Learning Engineer (September 2024-Present)
Involved with the Segmentation and Color recommendation part based on room background and furniture. Finetune YOLOv11-seg model with Custom dataset for Segmenting the Wall-Floor Ceiling and furniture. Used label-Studio tool locally for image annotation.
NybSys Inc.
Software Engineer (ML+Python Backend) (May 2022-June 2024)
Llama 3.2-3B LLM Fine-tuned with the custom company policy dataset. Used Langchain as framework for maintaining the agentic part and Prompt template, FAISS vector store for storing embeddings, and Sentence-transformers for retrieval. Used Fast-API for inference.
Pro Data Ltd.
Software Engineer (Android) (Mar 2021-April 2022)
Personal Projects
Retrieval Augmented Generation (RAG) Q-A on holy Quran(Translation) and Tafseer (Explaination)
Fine-tune Stable diffusion model with custom Dataset for image to image generation task
Medical data Fine-Tune model QA
Used Llama 3.2-3B for the Generative part. Used Langchain as framework for the Prompting, Referencing and Q-A, FAISS vector store for storing embeddings of Quran excel file, and Sentence transformers for retrieval of the verse.
This is a very simple (for learning) project. Used runwayml-stable-diffusion-v1-5 model for fine tuning with the custom sample fashion dataset.
Used a Text2Text Generation model (Bert based) for fine-tuning with the Medical Symptoms with disease dataset. Uploaded the model and hosted it as QnA using Gradio in Huggingface space.