Build Your Own Brain:* Lead the SaaS product's technology and data science initiatives.* Detailed information about the product can be found https://www.akaike.ai/byob-by-akaike.* Manage a multidisciplinary team of backend, frontend, and DevOps engineers to develop a robust, scalable, and extendable platform.* Oversee a team of data scientists to build generative AI pipelines, enabling advanced insight generation from various data sources and facilitating complex queries… Show more Build Your Own Brain:* Lead the SaaS product's technology and data science initiatives.* Detailed information about the product can be found https://www.akaike.ai/byob-by-akaike.* Manage a multidisciplinary team of backend, frontend, and DevOps engineers to develop a robust, scalable, and extendable platform.* Oversee a team of data scientists to build generative AI pipelines, enabling advanced insight generation from various data sources and facilitating complex queries surpassing the capabilities of existing chatbots.Infosec Lead & IT Infrastructure:* Oversee the company's IT infrastructure management.* Collaborate with HR and third-party IT vendors to establish optimal tooling and processes.* Achieved SOC 2 Type 2 certification for the product and the company, ensuring enhanced security compliance. Show less

Non-Contact Vitals Detection:* Led research to estimate heart rate, heart rate variability, and blood oxygen levels using video data.* Collected extensive video footage of faces and hands from a diverse demographic sample in various environments.* Developed a custom Fourier-transform-based video regression model and specialized loss functions, outperforming the accuracy of existing open-source models.AWS Billing and Monitoring Automation:* Designed and implemented a… Show more Non-Contact Vitals Detection:* Led research to estimate heart rate, heart rate variability, and blood oxygen levels using video data.* Collected extensive video footage of faces and hands from a diverse demographic sample in various environments.* Developed a custom Fourier-transform-based video regression model and specialized loss functions, outperforming the accuracy of existing open-source models.AWS Billing and Monitoring Automation:* Designed and implemented a monitoring system in AWS to actively oversee company resources.* Automated daily billing notifications and anomaly detection.* Identified underutilized resources, raising alerts to administrators and resource owners for proactive optimization.AI Healthcare Companion:* Developed a generative AI-based healthcare copilot.* Facilitates better understanding of medical conditions, analyzes medical reports, provides summaries of key points, explains medical terminology, and recommends next steps.Presales:* Actively involved in presales and sales initiatives.* Created project proposals based on client requirements.* Developed quick POCs to assess feasibility.* Led teams and oversaw projects as both Data Scientist and Technical Lead. Show less

Video Analytics:* Developed an end-to-end video analytics system for showroom environments.* Implemented a multi-camera tracking system capable of handling different camera types, including bullet and fisheye cameras.* Custom-trained models for accurate people detection in fisheye camera footage due to the lack of suitable open-source models.* Created a proprietary object tracking system for cameras with overlapping fields of view at various locations.* Designed demographic… Show more Video Analytics:* Developed an end-to-end video analytics system for showroom environments.* Implemented a multi-camera tracking system capable of handling different camera types, including bullet and fisheye cameras.* Custom-trained models for accurate people detection in fisheye camera footage due to the lack of suitable open-source models.* Created a proprietary object tracking system for cameras with overlapping fields of view at various locations.* Designed demographic detection models to ascertain age and gender of customers.* Developed customer classification models based on in-showroom behavior, despite the absence of uniforms on employees.* Built an analytical system to study customer behavior in showrooms, including product popularity and the impact of marketing activities on foot traffic.Recommendation Engine:* Engineered a recommendation system for software sales.* Built a solution to predict future software suite purchases based on the current software stack of customers.* Enabled sales representatives to target customers more effectively by leveraging patterns from other customers' purchasing behaviors.Automatic Banner Generation:* Developed an automatic banner generation system for a food platform, producing aesthetically pleasing offer templates.* Automated the creation of banners using input data such as restaurant name, details, banner food item, and offer CTA, adhering to predefined banner guidelines.* Reduced banner creation time from 20 minutes to just 10 seconds.Insurance Data Extraction:* Created a specialized model to accurately extract details from a variety of insurance cards.* Addressed the challenge of handling over 2000 unique card layouts, where existing open-source and commercial algorithms failed to deliver acceptable accuracy* Developed a semi-automated iterative MLOps pipeline to iteratively improve the model and facilitate labeling* Improved data extraction accuracy from 82% to 96% through these enhancements. Show less

Automated Brand Compliance:* Developed a computer vision solution to ensure showroom compliance with brand guidelines.* Tackled the challenge of limited negative examples by creating synthetic data, scraping the internet for additional data, and utilizing deep learning models like GANs to generate diverse negative examples.* Enhanced model accuracy from 60% to 95% through these innovative approaches.Rice Quality Assessment:* Designed a computer vision system to detect… Show more Automated Brand Compliance:* Developed a computer vision solution to ensure showroom compliance with brand guidelines.* Tackled the challenge of limited negative examples by creating synthetic data, scraping the internet for additional data, and utilizing deep learning models like GANs to generate diverse negative examples.* Enhanced model accuracy from 60% to 95% through these innovative approaches.Rice Quality Assessment:* Designed a computer vision system to detect defects in rice from a random 20g sample.* Achieved an accuracy rate of 98%.* Reduced the assessment time from 20 minutes to just 10 seconds.Assisted Radiograph Diagnosis:* Created a chest X-ray diagnostic tool to identify 10 different medical conditions (e.g., atelectasis, cardiomegaly, edema) and their potential locations.* Aimed to expedite the diagnostic process for radiologists, improving efficiency and accuracy.Aerial Image Segmentation:* Developed a computer vision solution for analyzing aerial images captured by drones at construction sites.* Provided comprehensive site analysis, including tree cover, buildings, lakes, elevation data, and land leveling estimates.* Reduced the analysis time from 30 minutes to just 20 seconds. Show less