Harsha Vardhan Tamatam Email & Phone Number

New York, United States

• Build vector stores for 15M+ users by processing cookies and transactional data using Langchain and LlamaIndex. Implemented advanced similarity search techniques on this data to enhance personalized marketing. • Developed a Python library to streamline data migration and ETL processes, with modules for a no-code interface simplifying SQL operations. Integrated unit tests and data validation modules to detect discrepancies and ensure accuracy.• Managed various data analysis tasks, A/B tests and handled ad-hoc requirements in a timely manner, ensuring project deadlines were met and stakeholders received actionable insights.

Bengaluru, Karnataka, India

• Developed a deep learning based recommended system to personalize video content on home page. Users’ click activity on the home page increased by 12%. Model Architecture: Candidate generation algorithms, Deep ranking techniques used• Developed statistical algorithms to optimize sales pipelines by utilizing users behavioral data and metadata we intelligently identify best users on the platform and their purchase intentions. Techniques Used: XG Boost classifiers, Db scan clustering• Built entire Data Engineering pipelines. Starting from, - Logging behavioral data on user platforms. Storing it on various databases and other analytics tools. - Building dashboards to extract and track various metrics and business KPIs. - Developing statistical and machine learning models to solve required use cases.

Gurugram, Haryana, India

• Worked on developing speech synthesis models. Focused on researching about speech models in English and other codemixed languages. Subsequently developed similar architectures for Hindi and other vernacular languages.• Built, trained and deployed text-to-speech models for Hindi, Telugu and Tamil languages. >100 hours of vernacular audio content is created using these speech models.

Hyderabad

I founded Autodub to help EdTech platforms, online educators, and video content creators expand their reach to foreign or vernacular audiences. With the rapid growth of online education and video content consumption, we automated the dubbing and translation of videos using cutting-edge AI techniques, combining technology and human intellect to deliver audio tracks in multiple languages at a fraction of the cost.- Developed an end-to-end automated speech translation pipeline, utilizing in-house deep learning models for speech recognition, machine translation, and speech synthesis.- Built a web platform with features such as proofreading and audio editing, enabling content creators to efficiently dub and translate their videos.- Translated and dubbed COVID-19 awareness videos into multiple Indian languages using our technology, gaining significant media attention and recognition.Visit: autodub.in

Masters Of Sciences, Data Science- Dept Of Statistics

Bachelor Of Technology, Computer Science Engineering