Jan 2021 – Nov 2024 · 3 yrs 10 mos · Mumbai, Maharashtra, India · Hybrid

• In IDFC, I had a chance to work on all the aspects of conversational bot lifecycle, starting from use-case selection and journey design to performance review and enhancement after go-live. Managed platform-infra, technical capabilities, development, integrations, project deliveries, dashboard and monitoring, performance enhancement and more. Currently managing Conv-AI capabilities as platform head, following are more details:
• Managing Conversational-AI platform capabilities, application server installations (with DevOps), channel integrations, bot development and integrations, APIs and backend system integrations.
• Performance improvement of the platform and bots across channels in terms of response time and infra utilization.
• Chat platform and bot integration with Telephony (Cisco), WhatsApp(Gup-Shup), Web and mobile net banking applications, IDFC website, and SFDC-chat for live chat agent transfer.
• Integration of Search-Assist, a ML based Website-FAQ-Crawler, with Service bot to improve coverage.
• Collaboration with management, business team, technical team, and vendors for project management, prioritization, effort estimations, and delivery. Vendor management (PS and OEM) to facilitate the bot-deliveries, and platform capabilities.
• Formation of Development and QA team, for development takeover from vendor professional-services and creation of testing strategy for functional testing of the bots.
• Functional flow creation with BAs for selected journeys for development and API identification/creation.
• Formation of Production Support Team for the platform and the bots, for L1 and L2 support. Implementation of the tracing for the backend API calls using open tracing and dashboards using grafana.
• Vulnerability Assessment, Penetration Testing, and AppSec closure with respective teams for all the integrations, application-servers and code-repos.
• Worked on Kore AI platform to create bots for both inbound and outbound.

Oct 2018 – Jan 2021 · 2 yrs 3 mos · Mumbai Metropolitan Region

• As a Senior Machine Learning Engineer lead team to deliver multiple client projects on production
• also developed new enhancements and capability for conversation bot and NLP. Followings are the few highlights:
• Analysing the historical customer logs, informative documents and creating Virtual Agent takes a lot of manual work. So our target is to automate the process as much as we can.
• I along with one SrMLE managed a team of 8 MLEs to create multiple enhancements like Intent Identification, FAQ gen, Intent Suggestion, Direct QA model etc which accelerated the process.
• To use all of them together as plugins, we create a platform with a sample Dialogflow agent in such a way that human interference was reduced significantly.
• The whole platform will work for some other Conversation Agent just by changing the dialogflow bot link(web-hook).
• This platform was used to showcase our capabilities to clients to get many breakthrough and projects.

May 2017 – Oct 2018 · 1 yr 5 mos · Pune/Pimpri-Chinchwad Area

• Ebex as marketing organisation wanted to analyse the textual information from the web and advertising headlines using NLP and deep learning.
• Researched possible solution of Automatic Headline Generation for given context(or previous headline) which would increase the conversion.
• Analysed headlines using feature vector (rule based) to create representation and used it to give scores

Jun 2014 – Apr 2017 · 2 yrs 10 mos · Pune/Pimpri-Chinchwad Area

• In Cognizant I worked on Java Development, ML, NLP and a bit of image processing.
• Created a platform to analyse Employee Performance using KPI metrics, using different scoring and analysis methods for Google Call Centre using Java as programming language and its open frameworks like Spring, Hibernate etc.
• Advocate Suggestive System to by analysing the judgement copies provided by High Court and suggesting advocate according to the case profile.
• QA System for Insurance queries to provide relevant paragraph to the query using bag of words and SVM