About Me
I am an experienced Data Scientist with a passion for Deep Learning. I like to solve real world problems with AI and have a proven research track record in Computer Vision and Natural Language Processing. Besides data science, I also like to watch anime and travel.
Discover a compilation of projects I've passionately developed during my leisure hours on this website.
Data Science & AI Research
I am a data scientist and AI researcher with 4 years of professional work experience and 3 years of academic research experience.
My journey in machine learning started with designing computer vision models for recognizing Indian Sign Language gestues on a live webcam video feed, implemented by using Haar Cascade Classifier. Having implemented this project at a school teaching Indian Sign Language in Mumbai, it felt rewarding and has been since motivated me to tackle real world problems using the power of AI. During my Masters Research, I took the challenge to design NLP model that would help people overcome Social Anxiety by suggesting sentences to talk about.
I love being up-to-date in the everchanging field of AI, by reading research papers about new algorithms and techniques. I try to implement the papers by applying the model architecture on a creative data to something unique.
Currently, I am applying my knowledge to solve problems in the healthcare industry tackling complex tasks involving health insurance claims, audits, and enable wide reach of medicines by promoting health insurance sales.
Projects
A collection of deep learning projects that I have created.
- All
- NLP
- Vision
- Generative AI
- Classical ML
- Visualizations
Work Experience
Professional Experience
Lead Data Scientist
April 2023 - Present
Cardinal Health, Chicago, IL
- Led a cross functional team of Data Scientists, Data Engineers, Software Engineers to implement multimodal email classification using Generative AI LLMs and Vector Databases to reduce costs by 40% and increase customer service productivity by 20%
- Pioneered prompt engineering strategies and conducted ethical AI assessments to establish guidelines for responsible use of Generative AI in patient and customer facing applications
- Designed and optimized prompts for Multimodal Generative AI models, significantly enhancing the quality and relevance of auto-generated responses in customer support scenarios leading to a 10x increase in customer service agent productivity
- Served as the Subject Matter Expert (SME) for Named Entity Recognition (NER) and extracted information using Document AI, Vision OCR. Designed NER models that achieved accuracy of over 94% while complying with HIPAA regulations
- Directed teams to develop automatic customer service email generations using 70+ billion parameter LLMs to reduce average agent case handling time by 60%
- Implemented Retrieval Augmented Generation (RAG) to efficiently generate customer service email replies, significantly improving agent productivity. Created dedicated knowledge bases using PGVector for generating emails for 50+ categories
- Conducted workshops and training sessions, to foster a data-driven culture within the organization, mentored junior data scientists, and contributed to the team's skill development and a collaborative work environment
Data Scientist IV
August 2021 - April 2023
Healthcare.com, Chicago, IL
- Designed Recommend Systems using Q-Learning Reinforcement Learning to optimize call buyer selection process and improve match rate by 11% uplifting revenue by $1.5 million per year
- Increased sales conversion rates by 19% by deploying Multi-Critera Decision Making (MCDM) Ranking Systems for call centers by statistically inferring sales agent performance
- Researched & implemented Natural Language Understanding (NLU) Speech Recognition and Transcription using TensorFlow ASR and AWS Transcribe to understand speech characteristics for sales and customer intent of buying
- Increased sales conversion rates by 8% using generative NLP Models, BERT, Transformers to produce sales scripts dynamically
- Examined Clustering (Machine Learning) Techniques in Python to identify characteristics of top performing agents
- Designed recruitment models that support in hiring top performing sales agents, successively adding $200k / year in revenue
- Created Ensemble Learning Models using Gradient Boosting Machines (GBM) and Synthetic Data to verify customer lifetime value (LTV) and accurately determine customer duration for health insurance policies
- Conducted training workshops on ML model training and CI / CD deployment using MLFlow, Kubernetes, GitHub Actions, AWS Lambda and Unit Testing for implementing standardized practices across the Data Science Team
Graduate Research Assistant (Thesis Research)
August 2020 - May 2021
Illinois Institute of Technology, Chicago, IL
- Researched & designed Generative AI virtual chat assistants that support humans with social anxiety to communicate efficiently
- Designed ensemble of open domain chatbots using GPT-2, Transformers and Encoders for context understanding and next sentence generation that helped to achieve fluent conversations
- Deployed Large Language Models into production on AWS EC2 using Flask, NodeJS, Express and VueJS for real-time predictions
- Improved Perplexity to 3.61 for testing data that was gathered while performing human study to achieve better conversations
Teaching Assistant
September 2020 - May 2021
University of Texas at Austin, Chicago, IL
- Provided critical training in Machine Learning & Data Analytics to a class of 60 students organized by Trilogy Educational Services
- Improved student proficiency in Data Gathering, Web Scraping, Pandas, Python, Tableau, SQL, Flask, D3 JS, Leaflet JS and React
- Conducting remote sessions and office hours to help students better understand Data Science concepts and Data Visualizations
Machine Learning Research Intern & Full Stack Developer
June 2017 - August 2019
Indian Institute of Technology, Mumbai, India
- Researched rapid object detection techniques for recognizing real-time video sign language gestures with Python and OpenCV
- Deployed gesture recognition models on web application for practicing sign language gestures with PHP, Python and JavaScript
- Identified key parameters that enhances the probability of students obtaining jobs at career fairs by analyzing student’s profile
- Obtained principal components by applying dimensionality reduction like PCA and t-SNE using Sci-Kit Learn, NumPy and Pandas
- Developed supervised and unsupervised ensemble learning strategies to predict pay range based on the candidate’s profile
Data Engineer & Full Stack Developer
June 2016 - August 2019
S & S InfoTech Services Pvt. Ltd., Mumbai, India
- Increased sales by 12% in four months by designing data warehouse and web application for managing eCommerce inventory
- Decreased 70% time to find 2.5 million profitable goods by analyzing supply chain through executing automated scripts and stored procedures using Microsoft SQL Server, ASP.Net and C# with Windows Server and IIS setup
- Reinforced banking industry’s email security by implementing DMARC policy to prevent email spoofing and domain abuse
- Provided geographical insights on email statistics and email spoofing attempts and generated analytical reports
- Prevented Data Leakage through building in-house web-based document management system to share files and reports with company’s Active Directory (AD) users and groups by Single Sign-On (SSO) authentication
- Improved company awareness and reach by creating context aware chatbot that provides company’s information using Google Cloud Platform and DialogFlow API and integrated it with Facebook Messenger and Skype
Educational Qualification
Master of Science & Computer Science
Illinois Institute of Technology, Chicago, IL
Thesis, Machine Learning, Natural Language Processing, Deep Learning, Combinatorial Optimization, Game Theory, Cloud Computing
Bachelor of Engineering & Computer Engineering
University of Mumbai, Mumbai, India
Artificial Intelligence, Human Machine Interaction, Distributed Databases, Analysis of Algorithms, Software Systems Architecture
Publications
Towards Assisting Human-Human Conversations
Diss. Illinois Institute of Technology, 2021 Thesis
DMARCBox - Corporate Email Security and Analytics using DMARC
5th International Conference for Convergence in Technology (I2CT). IEEE, 2019
FingerSpelling - Indian Sign Language Training Tool
18th International Conference on Advanced Learning Technologies (ICALT). IEEE, 2018
Exploratory Data Analysis using Dimension Reduction
IOSR J Eng (IOSRJEN) Best Paper Award
Improving Predictions Using Qualitative Parameters
JournalNX 3.08: 77-82
Recommendations
Stephen Perkins
Principal Data Scientist
I had the privilege of working alongside Tejas for over a year and I was consistently impressed by his dedication, hard work, and technical expertise. During our time together, Tejas made a significant impact at Healthcare.com by implementing and leading the development of an agent ranking system that improved our sales processes and lead conversion. His passion for innovation was evident in the creation of the call matchmaker recommendation engine, which used a Bayesian approach to match customers with the right buyers. This resulted in more efficient call routing and a better customer experience. In addition, Tejas was instrumental in identifying areas for improvement, such as his discovery that one of our proprietary insurance products’ LTV assumptions were 4x higher than reality. He also drove the adoption of best practices, including utilizing mlflow, Kubernetes, and unit testing, which helped to ensure the reliability and scalability of our systems. Overall, Tejas was a valuable member of our team and I have no doubt that they will bring the same level of commitment and excellence to their next role. I highly recommend Tejas for any opportunity in the field of technology and data science.
Carlos Zelada
Data Science Product Manager
I work with Tejas on several Data science projects. In the last project, we worked on a call routing algorithm where he increased a critical KPI. Tejas is a great data scientist with excellent technical knowledge. I was lucky to work with him, and he achieved great things. I'm sure he will add value to any company he is a part of. I can endorse him for his knowledge of AWS, python, MLflow, SQL, docker, lambda, and machine learning.
Kevin Park
Vice President, Product
Tejas is a brilliant problem solver. During my time at Healthcare.com, I greatly appreciated how Tejas was able to apply various advanced statistical and machine learning frameworks when tackling a new business initiative and explain in clear detail the why's and how's behind his approach. Tejas is very technically adept and effective at trying new tools and platforms for model development, testing, and deployment. He is an all-around team player and a rockstar with immense operating capacity, and any data science function would be fortunate to have him. I personally would definitely work with him again.
Prof. Shlomo Argamon
Chair, Department of Computer Science
Tejas has great technical abilities and communication skills, but more importantly, has tremendous energy and initiative - a great drive to build, solve, and succeed. Unlike most master's students, who take on thesis projects supplied by their advisors, Tejas approached me with his own original idea, and a great plan for developing and testing it. In the course of working with him on his thesis, he was consistently creative and hard-working, and also took criticism and suggestions very well. He will be an excellent addition to any team where his skills enable him to contribute, and I expect him to grow towards leadership very quickly.
Pratik Deshpande
SDE Prime Video
Attention to detail, and diving in deep to understand and solve the problem is something I learnt from Tejas. I've worked alongside him during my bachelors and developed Machine Learning and Full Stack Development projects. Tossing ideas back and forth and quickly analyzing the problem and patiently formulating a plan to tackle it, makes Tejas stand out. We share our passion for Machine Learning, and he is always excited to share his knowledge with anyone and is willing to help people reach their full potential. I am glad to have gotten a chance to work with him, and learn from him.
Prashant Mahajan
Amazon Search Science & AI
We rarely come across real talents who stand out like Tejas. I had the pleasure of working with Tejas for four years of undergraduate at the University of Mumbai, collaborating on several projects. Tejas' ability to learn new concepts and technologies quickly and explain it in a clear and concise way made a dramatic increase in the productivity level of our team. I am amazed by his diverse skillset across various domains of Computer Science, Machine Learning as well as a keen interest in research work. Tejas' talent as a researcher and fluency across numerous tech stacks made him a valuable member of our team. I recommend Tejas and I am sure he will be an asset to the team.
Rutvik Bhende
Data Science & ML Ops
I have known Tejas for many years, and have worked closely with him in many AI-based projects. Tejas is passionate about many things, and he makes sure his acumen in each is increased continuously. He is a devout learner and sympathetic team player. He is skilled at working on multiple tasks concurrently, without compromising the quality of the results. I always look forward to working with him again, in the future.