Highlights:

Skills:Deep learning, Machine Learning, Python, Research, Team management, Computer Vision, Natural Language Processing, Generative AI, Coaching and mentoring, Agile Development, PyTorch, TensorFlow, Lightening, Git, MLFlow, Steamlit and Gradio, Pandas, Numpy

Senior AI Researcher 05/2022 - current
CJ Express, Bangkok, Bangkok Thailand
Industry: Retail
In my previous role, I led diverse projects to enhance retail operations through data-driven solutions. This included developing a chatbot with advanced language models and optimizing Next-Basket Recommendation (NBR) algorithms for retail product bundling. I also contributed to in-store analytics, overseeing pipelines for person detection and image classification. Alongside technical work, I managed teams, fostered collaboration, and liaised with stakeholders. Through these efforts, I drove improvements in retail operations, revenue growth, and customer satisfaction.
• Led a cross-functional team of Data Engineers, Machine Learning Engineers, and AI Researchers in developing chatbot using LLM, RAG, LlamaIndex, and LangChain to support data team. - Explored advanced RAG techniques to improve groundedness and context relevance. - Established procedures for document enrichment and parsing to improve context relevance. - Investigated methods for generating and integrating knowledge graphs to enhance context retrieval. • Led research on Next-Basket Recommendation (NBR) using large-scale datasets(>1 Million data points). - Optimized TIFUKNN model for a large-scale dataset using FAISS. - Accelerated Sets2Sets model execution by 6x. - Conducted analysis of NBR results and built an algorithm for retail product bundling aimed at promotions and discounts. • Contributed to development of an in-store analytics solution to improve customer experience, and enhance productivity for retail outlets. - Built a pipeline to identify people and track people using person detection, multi-camera tracking, and Re-identification - Built an image classifier to distinguish customers from staff(>90% accuracy) - Use pipeline output and perspective transform to obtain store heatmap. • Gain team leadership and project management skills with 60-40 split between individual contribution and managerial workload. - Conduct regular one-on-one coaching sessions with team members. - Take initiatives within the research team to increase idea-sharing and cross-collaboration. - Engage with different stakeholders in backlog grooming and project timelines. • Explore diverse avenues of AI research and innovation like graph machine learning and diffusion models, contributing to the overall improvement of retail industry operations, increasing revenue, and fostering strong customer goodwill.--
AI Researcher 02/2020 - 04/2022
Sertis, Bangkok , Bangkok Thailand
Industry: Consultancy
In my role, I contributed to diverse computer vision solutions for clients across various industries. This included involvement in a License Plate Recognition project, creating a new Thai license plate dataset using GANs, implementing Bio-security measures for animal farms, and improving Personal Protection Equipment compliance. I also coordinated data annotations, developed an automated assessment system, and demonstrated expertise in machine learning problem-solving, contributing valuable insights to ongoing and future projects.
• Contributed to diverse computer vision solutions for clients across various industries: - Contributed to a License Plate Recognition project by training vehicle detection, and license plate classification models (>95% accuracy). - Created a new Thai license plate dataset utilizing Generative Adversarial Networks (GANs), facilitating enhanced diversity in data. - Implemented Bio-security measures for animal farm fine-tuning person detection (> 95% accuracy) and spatial constraints. - Enhanced Personal Protection Equipment compliance for hazard prevention by training person and equipment detection models(>95% accuracy). • Coordinate with a third-party vendor for data annotations for a large dataset annotation. - Created detailed documentation on data annotation - Constructed an automated assessment system for annotation quality using IOU-based metrics, accompanied by a PDF report generation. • Expertise in various aspects of machine learning problem-solving, such as data collection, cleaning, data and model versioning, and experiment tracking, ensuring accurate and reliable results. • Research and explore new avenues of research individually or in collaboration with teams, such as action recognition, people tracking, and acne detection.--
AI Research Intern 02/2019 - 07/2019
Elsevier, Amsterdam , Amsterdam Netherlands
Industry: Publication
In my role, I worked to identify diseases and map medical terms from articles using advanced NER and EL models, achieving stellar outcomes. Through multitasking, I streamlined model training, yielding a cutting-edge NER model. I explored Graph Neural Networks and diverse embedding techniques for EL enhancement. Additionally, I presented research at a key AI conference, contributing to medical text analysis knowledge.
Worked on identifying disease names in articles and mapping them to medical terms. - Developed models for Named Entity Recognition (NER, > 85% recall), and Entity Linking (EL, MRR 0.75). - Used Multi-tasking to train both models together obtaining SOTA NER model with 0.87 F1 in disease normalization. • Utilized Graph Neural Networks on disease graph data for normalization. - Trained various embedding techniques for EL, including Node2Vec (0.71 F1), Contextualized Node2Vec (0.72 F1), and Graph Convolutional Networks (0.71 F1). • Presented and published research as part of the master thesis at an AI conference--
Software Engineer 07/2014 - 05/2017
Unisys, Bangalore , Karnataka India
Industry: Software Development
In my role, I designed and developed software solutions, focusing on resource allocation for virtual machines and cloud-based disk drives. I improved VM availability and testing efficiency with innovative modules and tools. Mentoring our dynamic team of recent graduates was also a rewarding aspect of my role.
• Applied software engineering methodologies and techniques to design, develop, integrate, and qualify software solutions. - Contributed to development of modules for resource allocation, such as vCPU, memory and storage during virtual machine creation. - Improved the resource allocation queuing process, ensuring safe creation of VMs with high availability. - Contributed to development of cloud-based disk drive solution of magnetic tapes for a proprietary OS. • Contributed to development of a log analysis and monitoring tool increasing testing efficiency. • Mentoring team members of a young and dynamic team of mostly new collage graduates.--

National Institute of Technology, Silchar 08/2010 - 05/2014
Silchar, Assam, India
Degree: Bachelor's Degree
Major:Computer Science and Engineering
programming, algorithms, data structures, computer architecture, operating systems, database management, software engineering, networking, and artificial intelligence. The curriculum often includes theoretical courses along with practical labs and projects to provide students with a well-rounded understanding of computer science principles and their applications.

Login to view resume: Cv_May_8 -