Dec 2023 – Feb 2026 · 2 yrs 2 mos

• Leading a team of highly skilled engineers working on:
• > Voice-bot: Overtime we have built all dissected components of an AI powered voice-bot in-house. From ASR, to LLM, to TTS.
• > Voice Evaluation tool: Took initiative to have a framework for voice output quality evaluation, to scalably evaluate and weed out bad quality voice data for further improved training of voices. This tool has enabled creditas to train and build state of the art voice models.
• > Conversica (Voice Analytics tool): Made models to group insights from different calls on same accounts, drastically ruling out inconsistent facts in the conversations, yielding better accuracy of account priority scores for strategizing the right accounts.
• > Auto KYC kind of a micro platform to validate interview candidates background checks.
• > AI powered hiring tool, that takes interview call rounds for credit counselor roles, end to end automating the hiring process, without the need of a humongous HR team but only a couple HR folks.
• > Trained and helped deploy Speech to text model for conversica, this model parallels accuracy of industry best models for collection use case.
• > Initiated development of an evaluation framework for ASR, using different estimated methods of finding words, phrases, etc increasing the WER and their close matches.

Jan 2022 – Dec 2023 · 1 yr 11 mos

• Worked on several products and helped migrate a lot of them to more scalable systems:
• > In-house Campaign Management tool: Helped structure vendor integration seamlessly by implementing scalable coding and design practices.
• > Online digital collections portal: Standardizing each part of user journey, from login to post login subtle nudges to payments; to make a do-once highly generic kind of a replicable digital collections platform.
• > Digital journeys for several different banking products like: add-on, split card, savings account opening, etc.
• > Voice analytics tool: Built from scratch, an AI powered voice analytics tool; given input of an audio recording it can tell depths of collection probabilities and help strategize accounts for collection calling.

Apr 2021 – Jan 2022 · 9 mos

• Worked in reviews and rating team:
• > Trained and deployed a CNN from scratch to rank hotel images, most eye-appealing first.
• > Worked on Auto moderation of user uploaded contents: Images, Videos, etc.
• > Object detection model built for amenity check from uploaded images.
• > Crowd sourcing projects for different hotel attributes and amenities, nearby attractions, etc.
• > Built Video reviews platform from scratch.

Apr 2019 – Apr 2021 · 2 yrs

Oct 2018 – Apr 2021 · 2 yrs 6 mos · New Delhi Area, India

• > Developing Content optimization engine for websites
• > The product is being used on easeforgeeks.com

Mar 2018 – Sep 2018 · 6 mos · Mumbai Area, India

• > Worked with credit risk team for rural lending products using technologies: python, SAS and Microsoft Excel
• > Applied machine learning techniques like decision trees, Singular Value Decomposition, Mahalonabis Distance for various analyses.

Jun 2017 – Feb 2018 · 8 mos · Chandigarh Area, India

• > Developed several features for apps, including location based services(geofence specific) for business owner
• > Worked on nodejs, git, mongoDB, mySQL, linux, pm2, AWS to design and develop backend for TOOKAN, a business automation tool