San Jose, California, United States
Being able to design, engineer, deploy and manage complex AI software in production, together with teams of diverse, driven, diligent and dedicated individuals excites me. At IBM, I have delivered multiple AI/NLP capabilities (think algorithms, models, systems, and scale) to several different software products on-cloud, on-premise, and in deep collaboration with IBM Research and Software teams. These include scale-out execution of different models; algorithms and models to parse and understand legal language, complex tabular data, named entities and custom classes; systems to apply diverse combinations of NLP models on large collections of business documents towards structured and unstructured information retrieval etc., all the while working in and leading teams across global time zones and concerns. As a senior manager, I have also come to manage individuals, teams and their technical work, as they pursue their careers at IBM and deliver impact into its products and strategy. Prior to IBM, my graduate study at Stanford focused on Knowledge Representation and Reasoning, in the domain of natural language understanding and processing, using machine/deep learning, probabilistic and symbolic systems, a skill set that grounded (and continues to serve) me very well for my time at IBM and beyond.
Areas: Document Processing AI models and algorithms, Cloud/On-prem Back-end services Duties: Management of teams and individuals Products: WatsonX AI, WatsonX Orchestrate, Watson Discovery and Automation Document Processing - Manage global development and support: back-end AI services, library of AI models and algorithms, research to extract information from large volumes of complex business documents in PDF/Image/Office formats; programmatic, scanned documents of 10s to 1000s of pages with structures; perform multilingual OCR, table parsing, key-value pair extraction, structure analysis, page merging, trade-off resolution etc.; serves multiple products on-cloud and on-premise. - Manage local people/teams: 15-20 individuals across varying skill and seniority levels working on teams serving document understanding, document processing, natural language processing, DevOps and L3 support functions. Performance management, Hiring, Roadmap management, Agile project and release management.
Areas: Natural language processing AI, Cloud/On-prem Back-end services Duties: Staff to Senior-staff/Principal Engineering Products: BigInsights, Watson Compare and Comply, Watson Discovery, Weather Company - Led end-to-end research/development/evaluation/serving of NLP algorithms, models, services: --- parse and annotate complex tabular data like an expert analyst (eg. interpret financial reports) --- parse and classify diverse contractual agreements via semantic legal reasoning, like an expert attorney. --- recognize and classify relevant sentences, to augment out-of-box contract understanding. --- expand execution pipeline from 3 old models to 15 new out-of-box, custom multilingual NLP models. - Led API reviews across organization: Steered a global API committee in all the reviews and designs of all the internal, dark and public APIs published by Watson Discovery’s development organization (100-150 members) across all its teams and time zones. - Maintained Product Metering: Maintained back-end services and state management to measure instantaneous and sliding window usage of tenancy, features, resources, limits etc. across all product deployments and users towards billing and usage computations. - Applied in customer use-cases: Audit, compliance, procurement, agent-assist, financial analysis, RPA etc. - Mentored multiple junior engineers in their work streams and career aspirations.
Themes: Natural language processing AI, Distributed applications, Tooling Duties: Mid/Senior to Staff engineering Products: BigInsights, Watson Explorer - Developed, evaluated and enabled execution of AI models using SystemT: --- parse and classify sentiment polarities on multiple target mentions --- parse and extract named entity mentions, fundamental model used by most other models --- abstracted and scaled-out NLP model execution over very large data collections using distributed systems --- delivered tooling to enable different AI personas to develop and deploy NLP models over distributed systems - Applied in customer use-cases: Social-media and email analytics, Financial services, Marketing campaigns etc.
Assisted Prof. Jure Leskovec to teach/administer course CS224W. See attached media for course information.
Group - Oracle Linux & Virtualization Goal - Worked towards an intelligent algorithm for dynamic live migration of a virtual machine within a server pool assembly.
Served as a Technical Consultant under Prof.John Mitchell, to the Financial Services IT Consortium - A group of the IT divisions of certain financial institutions.