In my role as Technical Co-founder at moveme, I was instrumental in the ideation and execution of multiple projects, demonstrating the ability to successfully transition from feasibility studies to fully functional products. My responsibilities involved rigorous exploration and validation of innovative ideas, predominantly revolving around the application of deep learning models.Leveraging LSTM/GRU and Temporal Convolution Network methodologies, I created two distinct models for time-series classification and video dataset classification. These models utilized features extracted from advanced video models like VideoSwinTransformers, CLMR, and Trillson4, illustrating my proficiency in employing state-of-the-art technologies.Simultaneously, I delved into cutting-edge research to explore the integration of EEG data with videos for video prediction tasks, showcasing my continuous pursuit of innovation in the AI/ML field.Recognizing the importance of nurturing talent, I also undertook a mentorship role for a master's student, guiding her through a project focused on video classification.And finally, the development of an MVP, which is a unique product allowing users to search for movies based on their emotional preferences across various streaming providers (https://moveme.tv/). The promising feedback we've received and our robust positioning for future growth are testaments to the success of our efforts."

In my role at Tendium, I spearheaded the enhancement of a multi-label classifier model, a critical component of our product. I initiated a comprehensive strategy, starting with in-depth data analysis, understanding the labeling process, and scrutinizing the existing model. Recognizing the potential for improvement, I innovatively deconstructed the single, large model into a hierarchy of classification problems, applying a diverse range of models from logistic regression and decision trees to Roberta/transformers for the Swedish language, and fasttext.This strategic reconfiguration significantly boosted the precision and recall of each label, enabling fine-tuning of smaller label sets without compromising the system's overall performance. Additionally, I proposed improvements to the labeling process, ensuring accuracy and minimizing False Negative predictions. My contributions resulted in notable performance enhancements, reflecting my problem-solving skills and technical expertise.

As a Principal Research Engineer and founding team member, my multifaceted role included building, leading, and mentoring a dynamic team. I secured essential infrastructure, such as GPU machines, to handle a diverse portfolio of projects.Under my leadership, the team successfully delivered several key initiatives. These included a company-wide tool for textual data processing and classification, a PPE detection system for a mining company, a visual semantic search tool for a real estate firm, and a semantic segmentation model for identifying walls on a smartphone app for a paint company. Additionally, we developed a car make and model classification and a fine-grained car classification as internal learning projects, and guided an intern's final semester code search tool project.As an individual contributor, I helmed a range of projects, including a visual search tool for a hardware/tools vendor, a detection and search tool for a home decor company, and a text classification and entity extraction system for a legal firm. I also developed a time-series forecasting model for HVAC predictive maintenance for a real-estate builder in UAE and a company-wide document search tool. For internal research, I developed a Q&A model and conducted numerous experiments for uncertainty estimation in predictions.My technical expertise facilitated the successful implementation of research papers, the design, tweaking, and training of custom models in Computer Vision (Object Detection, Visual Semantic Similarity, Image Classification, Semantic Segmentation) and NLP (NER, QA, Classification, Retrieval, Re-ranking) using both deep learning and traditional machine learning. My contributions significantly enhanced our product offerings and customer satisfaction, demonstrating my innovative skills and value creation abilities.

• Primary responsibilities include doing research/study, design and development for the clients.• Currently leading a project for developing device driver of 3G USB modem device. The drivers will be built on Windows 2K/XP/Vista, Linux and Macintosh Tiger/Leopard. Apart from technical and project leading activities, developing a driver on Windows platform using KMDF.• Also leading a project for Communication API abstraction layer on Windows, Linux and Macintosh platform.• Designed and develop windows display driver and windows miniport hook driver, to have filter driver above windows graphics drivers.

Working with low-level and application level SW for implementing a convergence product combining the WCDMA mobile platform with laptop environment. Involved in designing and implementing low-level communication modules for communication between the PC and Mobile platform over USB. Responsible for writing validation logic as well as trouble shooting.

Design, development, and debugging of the user mode service, debugging and customizing the NDIS intermediate driver, and debugging RNDIS-related issues on the windows platform.

Interacting with customer. Debugging the VS2005, VC++ related issues. Analyzing crash dumps. Providing solution to customer.

Requirement analysis. Creating function specification, and design specification. Developing the software, customizing the system device driver.

Design, development, build, test, release and configuration management for several embedded and system side project including.1. Developing device driver for USB, PCI-Serial card on windows platform.2. Requirement analysis, designing, developing logic analyzer modules, designing framework for plug ins.3. Porting embedded OS on SPARC based processor.4. Writing an embedded driver for USB to communicate with client driverOther responsibilities are guiding and assisting colleagues in development.