Shruthi Chari, Phd Email & Phone Number

Lawrence Township, Nj, Us

Troy, Ny, Us

As a part of the IARPA HIATUS project:- Investigating representation of text explanations of ML outputs in authorship attribution setting using the Explanation Ontology (EO) and LLMs.- Developing a multi-step framework (decompose - delegate - synthesis) to go from user question (e.g., fine-tuning and generation from LLMs) to natural language explanation along Explanation Ontology (EO) supported explanation types.- In the delegate stage, directing the question interpretations to a suite of model explainer methods.- In the synthesis stage, extracting explanations from KGs using RAG techniques on foundation models (LLMs) (e.g., LLama2, Mistral)

Troy, Ny, Us

As a part of the IBM-RPI initiative Health Empowerment by Analytics, Learning and Semantics (HEALS), working on methods to support knowledge-enabled and user-centered explainability:- Designed an explanation ontology to structure and support the creation of different explanation types from the dependencies in the system, user and interface spaces. - Designed an end-end approach to provide contextual explanations to help clinicians better interpret machine learning predictions in a comorbidity risk prediction setting. Specifically, developedan end-end question-answering system that leverages large language models and knowledge augmentations to the same, to provide answers which contain contexts from authoritative medical literature like guidelines to better explain ML decisions like risk scores and model explanations of their scores.

I worked as an extern with the Biomedical Analytics and Modelling (BAM) team at the Center of Computational Health (CCH) at IBM Research. I worked on building a question answering approach to provide context in a risk prediction setting. I also collaborated with other students who were working on developing the clinical use case and risk prediction dashboard. This work is contributing towards my thesis research on the applications of user-centered explainability techniques in clinical use cases.

I worked as an intern in the Computational Computational Health Behavior and Decision Science (CHBDS) team and developed computational methods to model the lifestyle content of health users with an end goal of identifying behavior change literature that applies to them. During the internship, I interacted with a team of computer scientists, cognitive scientists and behavior scientists from the US as well as Europe. I presented my intern work during the end of summer poster session.

Troy, Ny, Us

During my time as a research assistant I undertook research to model and represent summarized statistics of study populations found in scientific literature. We developed a knowledge representation framework comprising of a study cohort ontology and knowledge graphs to represent cohort descriptions. Further, we powered cohort similarity visualizations and supported retrospective queries with the standardized knowledge graph representation. This work culminated in my Masters thesis, and has been submitted as an abstract and presented at multiple venues such as the Knowledge Graph Workshop at AMIA August 2019, US Semantic Technology Symposium (US2TS) March 2019, Automatic Information Data Discovery and Reuse (AIDR) May 2019, Northeast Computational Health Summit (NECHS) April 2019.

Troy, Ny, Us

Proctoring, grading and conducting labs for CSCI 1100 Introduction to Computer Science CS1.

Bangalore, Karnataka, In

I worked as a Software Developer on the various components of the search engine stack during my tenure as a Software Engineering intern at Cloudinfra. My tasks involved a fair amount of work in PHP, Python, Shell Scripting and AngularJS. -Remodelled the way in which search results were made available on the Expertrec extension available on PHP Ecommerce platform Magento.-Designed and implemented an NLP rule engine to parse, understand and map the price and colour attributes mentioned in a search query term.-Added support to boost certain fields and suppress certain fields of a searchable entity, to achieve fine tuning of search parameters. Designed the AngularJS front end component for the same.-Automated the test suite for the magento extension Expertrec using Docker. Wrote pre-build and post build test scripts for the same, this considerably reduced the deployment and build time of a release.-Built a generic framework to generate custom autosuggestions by giving flexibility to choose the search attributes and filler texts that need to be shown on search suggestions. -Worked on a portion to synchronise phrase and synonym addition, when synonyms were added/edited on the front end by the customer. The multi word phrases contained in the synonym were recognised and added to phrases, to boost Solr's search performance.

Santa Clara, California, Us

I continued to work as a software developer in the NOS Engineering team, on the internal provisioning tool ANMC. I designed and implemented larger features which required me to design more hibernate models and design the SQL schema to map the modelling, and I also graduated to wholly design UI workflows to accommodate my backend requirement needs.-Worked as a lead developer on the resource modelling project. Remodelled the resource computation calculations to accommodate the limits of a machine's connected sites(nodes); instead of a global network entity based computation which solely considered a machine's resource limits alone. This increased the accuracy of the resource calculations ~5%.-Added support for new virtual network interface(VLAN) connections to be provisioned for network devices, the addition of support for these devices required refactoring the existing hibernate and database modelling. -Provided a Grails interface to commission new DHCP relay servers at the customer's site level. The task of writing algorithms in Java and Groovy, to offer IP addresses from the floating DHCP pool, based on the subnet and ranges specified proved to be challenging.

Santa Clara, California, Us

I worked for the NOS-Engineering team at Aryaka, Bangalore. I was a full stack developer working on the Operator Provisioning Portal, this portal is used by the Support Team to serve the requirements of our customers. -Designed a system to synchronize updates to a configuration file that is changed by two entities. This was done using Timer Tasks in Java.-Provided support to add multiple SSL certificates to the same ip/port combination for a customer. This involved reworking the restrictions on the existing framework.-Worked on the existing framework and data stores to rename and merge certain entities. ETL scripts were provided for the same, and code cleanup was performed on the existing Java and Grails codebase.

San Jose, Ca, Us

I interned at the Cisco IT center in Bangalore; and I worked in the WebeX IT team, during this brief stint. I worked on re-engineering a small portion of the SaaS billing middleware. The existing Account Management codebase was in simple Java, and hence to make it more flexible and modular ,and to be able to support future business changes I ported the code to Spring and Apache chain. Apache Chain allowed to specify the flow of pattern for the creation of multiple accounts in the scenario of supporting various account types and at multiple tiers like that of partners, reseller and end customer. I was also developing content for a team poster, describing the ongoing projects and the architecture of the iServices middleware framework.

I served as a research intern at the Knowledge Analytics and Ontological Engineering Center(KAnOE) at PESIT, Bangalore. My area of research encompassed semantic web and linked data. Developed techniques to retrieve, integrate and build Linked Datasets from readily accessible data stores. Worked on ways to use Semantic Web concepts to make this data easily accessible to the layman. I co-authored and published two posters to the ISWC, during my stint here.

Santa Clara, California, Us

I worked on a resource estimation project during my internship at Aryaka. This project added a resource estimation feature to the Operator-User ANMC portal used in Aryaka, modelling a scenario that determines the feasibility of the addition of sites at various levels in the hierarchy of the core network infrastructure.

Doctor Of Philosophy - Phd, Computer Science

Master'S Degree, Computer Science

Engineer’S Degree, Computer Science

Icse