Oct 2022 – Feb 2024 · 1 yr 4 mos · Bengaluru, Karnataka, India · Hybrid

• Discovery Platform Infrastructure: Spearheaded the end-to-end development of a robust discovery platform infrastructure, integrating 30+ filters and creator reports. Implemented real-time data retrieval mechanisms from diverse sources, including third-party data aggregators and also platforms such as YouTube (Data APIs) and Meta (Graph API queries), ensuring up-to-the-minute data accuracy.
• Demonstrated deep product expertise by rapidly conceiving and executing a Creator report, encompassing an extensive array of 100+ metrics, using real-time data. Optimized performance through parallel processing techniques and introduced PDF report generation, necessitating the implementation of queue management and email functionality. OAuth2 framework to enable
• access to Private Metrics of YouTube with varying scopes, enhancing data access and security were also added.
• Designed and deployed a dynamic subscription model for users of the discovery platform, facilitating the acquisition of credits. Achieved swift deployment, enabling the onboarding of initial customers, and subsequently expanded the model to include usage history tracking at a user level. Attribute-based access control was integrated into the system.

Jul 2021 – Jul 2022 · 1 yr

• Built an NER-based NLP pipeline for Medical Record Transcriptions. Partnered with Clinicians to develop the parameters required for diagnosis. The module has an intuitive visualizer, which is a one-stop point for clinicians to assess years’ worth of patient records easily, making diagnosis faster, saving time in case of emergencies.
• Open Telemetry: Tracing implementation for the Services Platform. Open Telemetry implementation for python micro-services, with common collector and visualization in jaeger and zipkin. This helps developers understand multi-layered architecture, making it easy to navigate from effect to cause in a production system.
• Golang based microservice for Edison Services Platform. Enterprise level implementation with the ability for older products to talk to the central application through a communication layer. This will save approximately $100k dollars per year, with service teams requiring lesser number of visits for updating software to scanner products.

Nov 2019 – Jun 2021 · 1 yr 7 mos

• The Edison Engineering Development Program (EEDP) is a 100+ year-old initiative and is a 2-year rotational development program designed to accelerate learning, with business-critical assignments and broad exposure to engineering.
• Rotation #1 & #2:
• Built a DNN-Regressor based Recommendation Engine to reduce scan time on MRI scanner. This helped in simplifying workflow in the prescription phase, reduce scan time by up to 50%, and increase patient throughput in the patient space, with minimized rescans. Almost 100k datapoints were deanonymized and cleaned for training. This won GE MR Innovation Competition, in the application category, held in Milwaukee, USA.
• Integration of machine learning module with Java Microservice Platform. In the Scanner Platform, recommendation engine was integrated with multiple other services. Additional parameters were required in the processing with range estimation based on dynamic change of Nuclear Medicine Parameters.
• Rotation #3:
• Built a Mask-RCNN based tool for patient BMI and gender prediction. This tool simplifies protocol selection, with multiple other use-cases across the patient space. Collaborated with multiple product managers to perform cohort analysis that identified this opportunity. Clinical Domain Knowledge and HIPAA are important learnings from this project. This included an additional layer for nudity detection, with real-time data input from intel RealSense camera.

Nov 2018 – May 2019 · 6 mos

• Decentralized Healthcare, built on Blockchain. Patient data management, Secure data storage and interoperability were the priorities.
• Designed and developed an MVP of a medical platform on Blockchain. Mentored in IIM Bangalore. The vision of the company was to decentralize medical data and make it accessible to individuals.
• Developed communication infrastructure for patient data, multiple web applications across different vendors. Hyperledger Indy was the network used in the development, and this distributed ledger was purpose built for identity. Zero Knowledge Proof which doesn’t require any additional information was used in the system.

Jul 2018 – Sep 2018 · 2 mos · Hazira

• Ground-Level knowledge of working on Petrochemical Plants. Underwent extensive training in Environment, Health and Safety. Developed POC for fault analysis using Machine Learning algorithms like RNN.

May 2017 – Jul 2017 · 2 mos · Hazira

Jan 2017 – Dec 2017 · 11 mos

Jan 2016 – Dec 2016 · 11 mos

Dec 2014 – Feb 2018 · 3 yrs 2 mos