EZOFIS, under Seth's leadership as CEO for over six years, delivers innovative low-code AI workflow automation solutions tailored to enhance business productivity and efficiency. The company focuses on streamlining operations, eliminating inefficiencies, and driving profitability by leveraging advanced methodologies and cutting-edge technology.  

With multiple certifications, including Lean Six Sigma Green Belt and expertise in business growth and strategy, he prioritize empowering organizations to modernize processes and achieve measurable results. His mission is to enable seamless digital transformation for clients, fostering sustainable success in a competitive market.

Seth ™

Laivly

- Spearheaded the development and deployment of machine learning models to tackle complex business challenges, leveraging expertise in experimental design and demonstrating meticulous attention to detail.

- Engineered scalable, end-to-end machine learning solutions with a strong focus on performance optimization and operational efficiency.

- Led critical project implementations, driving impactful results through effective collaboration with cross-functional teams and showcasing leadership in delivering ML-driven innovations.
Played a pivotal role in evaluating cutting-edge technologies and contributing valuable insights during technical solution discussions, ensuring alignment with business goals.

- Drove significant improvements in efficiency and accuracy by leading the creation of machine learning solutions tailored to address specific organizational needs.

- Ensured data readiness and reliability by managing comprehensive data cleaning and preprocessing tasks, highlighting exceptional data science and analytical skills.

- Proactively monitored and maintained machine learning systems, identifying potential issues early and implementing solutions to enhance system functionality and performance.

- Explored alternative problem-solving approaches and extended technical capabilities to support seamless project implementation, ensuring adaptability and innovation in solution delivery.

Data Scientist

EZOFIS

AI Engineer

University of Toronto

CSC148H1S: Introduction to Computer Science (Summer 2023, Winter 2024)

Graduate Teaching Assistant

SOTI

- Conducted intricate Lithium-ion battery experiments, revealing interrelationships among parameters. Employed 4+ ML models (ARIMA, SARIMA, etc.), alongside 'Coulomb Counting' and 'Entropy Calculations', to precisely determine battery State of Health (SOH).

- Enabled the seamless integration of non-smart batteries into SOTI's XSight dashboard by deploying ML models on Android devices, resulting in a positive impact on over 80% of SOTI's existing clients.

- Designed and implemented an Android application to assess battery health, employing a tailored Coulomb Counting method, achieving an impressive accuracy rate of 90%. The application is now slated for deployment across all SOTI customer environments.

- Developed a novel indoor localization system leveraging smartphone capabilities, effectively utilizing Wi-Fi signals and magnetic field data, eliminating the need for additional hardware. The resulting machine learning model achieved an accuracy rate of approximately 82%, significantly reducing positioning errors to within 4 meters.

Research & Development (Intern)

16 Bit Inc.

- Spearheaded an ambitious project leveraging Large Language Models to generate Medical Reports from X-Ray images.

- Designed and implemented a transformer-based architecture, enabling efficient feature extraction from multiple X-ray images per patient. These features were then fed into a decoder to generate detailed findings and impressions.

- Curated and anonymized an extensive dataset of 50,000+ medical images and reports which were stored in an SQLite database that served as the foundation for model building and research.

 - Ran several parallelized experiments showcasing the project's immense potential to streamline and optimize the medical reporting generation process.

DeepTek

- Modeled a task for segmenting and detecting foreign bodies leveraging the UNET architecture and the Xception backbone, which is now being evaluated by the FDA for approval. Including foreign objects such as pacemakers, sternal sutures, endotracheal tubes, cardiac valves, chest leads, etc., the model achieved a sensitivity of 95%.

- Designed and implemented a cascaded U-net segmentation and classification model for detection of an endotracheal tube in a given radiograph, identifying the position and determining whether it is inserted correctly or not

- Successfully implemented segmentation models for pathologies like edema, and sternal sutures and for classification of normal vs abnormal X-RAYs. 

- Curated, annotated, reviewed, and imported 5000+ radiographs in the MongoDB database which was used across multiple teams for model building and research.

- Initiated and successfully implemented a new workflow for radiograph annotation and review process for the model building and experimentation which saved 5+ working hours per week for the organization.

- Performed topic modeling using the techniques in Natural Language Processing to predict the type of modality from the text corpus and comments provided by the doctors in the patient reports with 98% accuracy.

Associate Data Scientist 

ACI Worldwide

- Worked on the development of multiple microservices using .NET core and C# to upgrade existing features and to add/ incorporate new features for the Speedpay product.

- Programmed a feature, ‘Agent Card Assist’ using ASP.NET Core for real-time account lookups to accelerate loading and authentication of customer’s user accounts that helped to retain critical clients.

- Contributed to the integration of ApplePay and GooglePay in the Speedpay product across multiple payment channels such as Extranet, Internet, IVR, etc for a seamless payment experience to the client.

- Presented an idea to optimize the organization's cross-team communication platform in the "Ideas in Action" event which received the "Most Engaging Idea" award.

Associate Software Engineer

Excel Switchgear Services

- Engineered a cutting-edge, responsive chatbot using NLP and ML, transforming customer interactions and operational efficiency.

- Implemented robust data collection and preprocessing, ensuring the chatbot delivers contextually accurate and personalized responses.

- Crafted a chatbot that engages users proactively, anticipating needs through predictive analytics and intelligent algorithms. 

- Showcased tangible results, quantifying the chatbot's success in improving customer satisfaction, reducing query resolution times, and enhancing overall operational efficiency.

Intern

Persistent Systems

Trainee

Devashish is a recent graduate of the University of Toronto, where he earned a Master’s degree in Applied Computing. Throughout his academic journey, he was honored with prestigious scholarships, including the Vector Scholar in AI and the JN Tata scholarships, for his outstanding academic performance. With a solid foundation built over four years of experience in software development and as a Machine Learning Engineer, Devashish brings a unique blend of research acumen and technical expertise.

During his tenure at SOTI.inc, Devashish delved into machine learning and mathematical modeling, focusing on optimizing and predicting lithium-ion battery health. His Research and Development Intern role gave him valuable insights and practical experience in this specialized domain. Before this, he served as a Data Scientist at 16bit, where he curated image and textual data and implemented multi-modal AI models to generate radiology reports using DICOMs and clinical comments leveraging Large Language Models (LLMs).

Devashish's contributions at DeepTek.ai were marked by his proficiency in developing FDA-approved chest X-ray segmentation models for various pathologies, showcasing his prowess in PyTorch, computer vision, image, and language processing domains. Furthermore, his tenure at ACI Worldwide as a Software Engineer equipped him with in-depth FinTech domain knowledge and honed his software engineering skills, including working with large databases, C#, .NET, GitHub, APIs, and Cloud.

Devashish is eager to leverage his diverse skill set and rich experiences to make meaningful contributions to his future endeavors, and he is confident that his profile aligns seamlessly with the requirements of the software engineering, ML Engineering, and data science roles he aspires to undertake.

taplio

The Battle

Seth ™

Devashish Khairnar

Followers

Posts

Engagements

Likes

Ready to grow your LinkedIn presence?