Abhijit Ramesh Email & Phone Number

Santa Cruz, Ca, Us

New York, United States

At Theta Tech, we develop medical AI software for clinicians, researchers, academics, and small businesses. As a machine learning engineer, I work on all parts of the lifecycle of this software. The client projects that I am involved with are:QuantaFlo, Semler Scientific: - Developed the current deep learning model for diagnosis of Peripheral Arterial Disease. - Build a highly scalable Extract Transform Load pipeline with AWS. - Build an auto-labelling tool for doctors to label good and bad quality data.LungRadCad - Center of Computational Imaging and Personalized Diagnostics (CCIPD) at Case Western Reserve University - Worked on an uploader where doctors and clinicians can upload their research data with the LungRadCad OHIF viewer. - Research and Develop a content-based image retrieval system using Siamese Networks. - I have implemented a user management panel to add doctors, researchers, and clinicians to the LungRadCad system.Patch Sorter - Andrew JanowczykA tool for rapidly labelling objects using deep learning feature embedding. - I help Andrew fix bugs and implement new features for the patch sorter open source tool.

Kochi, Kerala, India

TexNano works directly with Doctors to develop AI solutions for their research. We are currently working very closely with the amrita institute of medical sciences. Alongside Dr. Priya Nair and Professor Gilad Gressel, I am researching deep learning methodologies for the early detection of Pancreatic Cancer. I have worked on this project under the same mentors since my senior year in college. This project was further taken up under TexNano on September 2022 after realizing its potential in the real world. We are currently working towards its implementation at Amrita Hospitals to assist the endoscopist in quickly recognizing tumors in the pancreas and identifying the best biopsy zones.

Remote

Mentored Suchith Krishna S Donni and Ziwei Cao on Phase 5 of the machine learning scorecard for credit risk assessment. This year we integrated an end-to-end federated learning pipeline using Pysyft, Flower and FATE. This allowed Mifos to develop credit risk models using federated learning from different financial institutions while ensuring data privacy.

My major contribution to the Mifos initiative was as a mentor for the Machine Learning Score Card Project, which helps financial institutions with credit risk management. I have also worked on their client-side application mifos mobile cloud network as part of google summer of code 2019.

Remote

I worked as a mentor for the project mifos-mobile-cn; the primary goal of this year's GSoC was to implement Open Banking API to provide secure authentication and integration layer for first-party applications.

Worked with Professor David Lee at Tech4Good lab on the research and development of an AI approach to investigate and support apprenticeship learning in the economy. The key area of research here involves creating a reinforcement learning system that mimics economic situations, which involves mimicking the real-world professional context of delivering projects. The agent would be provided with opportunities or guidance, support as well as receive critique on the completion of the projects, which means modelling the complex interactions and dynamics involved in a purely analytical model.

Tenkasi, Tamil Nadu, India

Worked with Dr Jayaraj Poroor - Principal Researcher @ Zoho Corporation, on improving the current data pipelines of the company by researching methods to integrate domain knowledge into deep learning models. I developed neuro symbolic methods to synthesize data from existing mathematical models that the domain experts have already proposed; this worked in tangent to data learning the deep learning model already has, thereby improving the learning exponentially. The proposed learning method showed a significant improvement in accuracy to existing data models in Zoho Sheet and other products currently in the making.

Kollam, Kerala, India

amFOSS is a student driven community in Amrita Vishwa Vidyapeetham, Amritapuri which helps mentor students in different field in computer science technically. The main focus of the club is to encourage open source contributions to different organisations. Students also actively participate in contribution drives, hackathons, research and competitive coding.

Remote

Mentored Nasser Yemdjih Kaze to develop phase 4 of the machine learning credit scorecard. This project provides financial institutions with a tool to minimize credit risk. We built RESTful APIs using Django around existing models for credit risk assessment during GSoC 2020. We deployed the project to the fineract platform by developing front-end and backend integrations in the Mifos web app and Fineract 1.x backend.

Mifos-mobile-cn is an Android app for digital financial services built on top of Apache's Fineract Platform. I worked on improving the application's user experience, functionality, and overall aesthetics. I integrated the following features: - Support for customers to add their details - Support for financial institutions to review the loan request - Show recent transactions on the application's homepage - Integration to show loans and other financial productsSome essential core android tasks that I did are: - Migration of the whole project to Android X - Integrate Travis CI built - Code cleanup and warning fixes

I have actively contributed to the fineract-cn-mobile project of the Apache Software Foundation.