Aug 2024 – Mar 2025 · 7 mos · Boston, Massachusetts, United States · Remote

• In my role as a Senior Software Engineer at a Stealth Startup in Boston, Massachusetts, I spearheaded the development of a GenAI based approach for marketing campaigns. By designing and implementing experiments to evaluate LLMs for messaging frameworks, I was able to enhance ad content across various platforms. Additionally, I developed a creative reasoning framework to inject emotions into ad content, ultimately improving engagement and conversion rates.

Apr 2021 – Aug 2024 · 3 yrs 4 mos

• Worked on the productionization of Generative AI on Twilio Segment data
• Developed a framework to formulate and validate system prompts for the Generative Audiences Usecase within Twilio Segment
• Developed a standardized unit-testing framework to evaluate the performance of LLMs and use it to iteratively improve the system prompt for the generative audience use-case
• Development of a Proof of Concept usage of vectorDB to pass on only relevant context to the system prompt for LLMs for Generative Audiences usecases and beyond.
• Predictive Traits
• Developed intial proof-of-concept approaches of computing the life-time value of customers for predictive traits within Twilio.
• Worked on productionizing proof of concept and made the product usable for general availability.
• Identify and implement best practices for unit-testing framework code to ensure reliable CI/CD

Jun 2019 – Apr 2021 · 1 yr 10 mos

• Lead on development of speech recognition models for our ASR platform.
• Lead on the development of manual data asset acquisition pipelines for speech recognition.
• Focus on development and application of hyper-parameter optimization techniques for optimizing towards multiple objectives for speech recognition models.
• Lead on development of evalution framework for enabling comparision of speech recognition performance across multiple metrics across vendors.

Jun 2017 – Jun 2019 · 2 yrs · Belmont, CA

• Developed data pipelines, speech recognition models and optimization of deep neural networks for multiple languages.
• Development of a continuous model update framework for multiple languages.
• Contributed to the development of dynamic word addition capability to our speech recognition engine.
• Exploratory work into speaker identification and speaker diarization for real-time speech recognition systems.

May 2016 – Aug 2016 · 3 mos · Sunnyvale, CA

• Worked on importance sampling-based data sampling techniques to improve training time for speech recognition.
• Worked on improving the speech recognition toolkit for better visualization of results.
• Implemented a framework to allow different data selection strategies for deep learning.

Oct 2013 – May 2014 · 7 mos · San Jose

• Working on Multimodal Interaction Manager for Mobile Environments.
• Exploration of machine learning algorithms for domain classification of spoken sentences.
• Co-inventor in 3 Patent applications resulting from this work:
• Multi-modal fusion engine
• Selecting Multimodal elements
• Identification of user input within an application

Jan 2012 – Dec 2019 · 7 yrs 11 mos

• I was a Ph.D. Candidate in ECE. My research is focussed on
• Research and development of hyper-parameter optimization strategies to build deep neural network (DNN) systems for speech recognition on embedded platforms.
• Methods to utilize learning curves from previously completed experiments to predict the performance of a new system build.
• Rapid prototyping and evaluation of speech recognition engines for various languages in the US Govt. funded multimillion dollar project BABEL.
• Data annotation and analysis of human interaction in a multi sensored car environment as a part of the Honda Research Institute sponsored CESAR initiative.
• I have worked on acoustic fingerprinting techniques for music retrieval to perform monitoring and matching of radio data for targeted advertisement delivery and song recognition.
• Accomplishments
• Authored a paper on optimization of decoder hyper-parameters for ASR at IWSDS 2016.
• Worked on the implementation of deep neural network based acoustic model computation on OpenCL enabled android devices.
• Co-authored a paper on the data analysis from the CESAR project.
• Implemented and accelerated frame level lattice generation for existing Janus Speech Recognition Engine.
• Built and evaluated speech recognition systems for low resource languages including Cantonese, Pashto, Turkish, Tagalog.

Jan 2011 – Dec 2011 · 11 mos

• Worked on The Imagined Speech using EEG Project

Sep 2010 – Dec 2010 · 3 mos

• Worked on the Axonal Bouton Detection Project

May 2009 – Jul 2009 · 2 mos

• Designed and implemented an SMD PCB Reflow Oven Alarm system.
• Designed and implemented a factory-wide multi-programmable alarm with RTC.