Setting up the general platform and infrastructure to help Data Scientists deploy new use cases faster

Develop and evaluate complex agent use cases like helping patients with questions about Financial Aid applications

Led the development of a feature to display Health Benefit Accounts (HBA) to patients, requiring integration with multiple HBA vendors. Designed a scalable solution adaptable to future vendor integrations, ensuring long-term maintainability.

Spearheading the creation of a Configuration Management app to streamline product setup and configuration. This app supports provider setup and facilitates importing/exporting configurations across environments, significantly simplifying environment setup

Collaborated closely with stakeholders to define requirements, align on technical approaches, and deliver solutions that enhance the patient experience and operational efficiency.

Mentored team members, fostering a collaborative and high-performing development culture.

Leading Arthur Shield, providing real-time protection for LLMs, from conceptualization to successful POC implementation.

Pioneered the implementation of robust authentication and authorization using Keycloak, a major undertaking requiring extensive research and understanding of authentication systems, OAuth, and various providers.

Collaborating with ML Research team to integrate cutting-edge techniques for hallucination detection.

Developing configurable APIs for Shield, empowering users to tailor the product to their unique use cases.

Spearheaded the creation of a demo chat application using RAG and Shield APIs, showcasing product capabilities to potential clients.

Mentored junior engineers and ensured adherence to engineering best practices.

Developed asynchronous workflows for model onboarding and led the design and implementation of an integration testing framework.

Led design and architecture efforts for scalable, reliable, and robust products.

Mentored junior engineers by reviewing their work, and providing guidance for good engineering practices and code design.

“Virtual Scribe” is a passive listening service for care delivery, designed to reduce the cognitive load and fatigue of physicians and care providers. Virtual Scribe audio records patient-physician conversations, transcribes speech to text, extracts cognitive clinical concepts, and codes key terms using natural language processing.

My Responsibilities: Led backend design and development effort, developed services using microservices and event driven architecture, collaborate on component-level technical designs, participate in code reviews, provide release support. Drove the effort in setting up Kinesis data streams and serverless Lambda functions to process streaming audio and text data that streamlined the use of the data by multiple consumers. Cloud deployment of services using Amazon ECS, set up CentralizedConfiguration management.

Impact: Reduced the down time during deployment to virtually none by utilizing Blue GreenDeployment and AuotScaling. Integrated the service with New Relic to monitor the performance and identify bottlenecks in real time.

Technologies used: Java, Spring Boot, Restful Web Services, Python, javaScript, Node JS, AWS(ECS, EC2, ECR, S3, Lambda, Kinesis, Medical Transcribe, Cloudformation, Code Deploy, MQ,Secrets Manger, ELB, Route 53), Redis, Splunk, MySQL, Docker, Maven, Jenkins.