Abhinav Chinta

Machine Learning Engineer at Typeface
  • Claim this Profile
Contact Information
us****@****om
(386) 825-5501
Location
Sunnyvale, California, United States, US

Topline Score

Topline score feature will be out soon.

Bio

Generated by
Topline AI

You need to have a working account to view this content.
Join now
You need to have a working account to view this content.
Join now

Credentials

  • Neural Networks and Deep Learning
    Coursera
    Apr, 2019
    - Nov, 2024
  • Applied Social Network Analysis in Python
    Coursera
    Nov, 2018
    - Nov, 2024
  • Fundamentals of Digital Image and Video Processing
    Coursera
    Apr, 2018
    - Nov, 2024
  • Software Processes and Agile Practices
    Coursera
    Apr, 2018
    - Nov, 2024

Experience

  • Typeface
    • United States
    • Software Development
    • 1 - 100 Employee
    • Machine Learning Engineer
      • Aug 2023 - Present

      Developing Generative AI solutions to accelerate content creation for brands. Developing Generative AI solutions to accelerate content creation for brands.

  • Amazon
    • United States
    • Software Development
    • 700 & Above Employee
    • Data Scientist
      • Jun 2022 - Jul 2023

      Amazon SageMaker Ground Truth - *Delivered tailored synthetic images to customers to improve their downstream model performance using different AI-based approaches ranging from color transfer to generative models (specifically CycleGAN for style transfer and Stable Diffusion for infilling). *Successfully demonstrated the value of synthetic data across different modalities using generative models and defined metrics to quantify the domain gap between synthetic and real… Show more Amazon SageMaker Ground Truth - *Delivered tailored synthetic images to customers to improve their downstream model performance using different AI-based approaches ranging from color transfer to generative models (specifically CycleGAN for style transfer and Stable Diffusion for infilling). *Successfully demonstrated the value of synthetic data across different modalities using generative models and defined metrics to quantify the domain gap between synthetic and real data. *Constantly experimenting with latest developments in the field of generative AI to deliver synthetic data solutions for a wide range of customer needs. Show less Amazon SageMaker Ground Truth - *Delivered tailored synthetic images to customers to improve their downstream model performance using different AI-based approaches ranging from color transfer to generative models (specifically CycleGAN for style transfer and Stable Diffusion for infilling). *Successfully demonstrated the value of synthetic data across different modalities using generative models and defined metrics to quantify the domain gap between synthetic and real… Show more Amazon SageMaker Ground Truth - *Delivered tailored synthetic images to customers to improve their downstream model performance using different AI-based approaches ranging from color transfer to generative models (specifically CycleGAN for style transfer and Stable Diffusion for infilling). *Successfully demonstrated the value of synthetic data across different modalities using generative models and defined metrics to quantify the domain gap between synthetic and real data. *Constantly experimenting with latest developments in the field of generative AI to deliver synthetic data solutions for a wide range of customer needs. Show less

  • The Johns Hopkins University
    • United States
    • Higher Education
    • 700 & Above Employee
    • Graduate Research Assistant - Center for Language and Speech Processing
      • Jan 2021 - May 2022

      I successfully fine-tuned an LLM (specifically BART) to improve text summarization results for my capstone project under the supervision of Prof. Mark Dredze. My area of research was the summarization of Twitter tweets discussing specific topics (or hashtags) using an extractive/abstractive summarization approach. Previously under the supervision of Prof. Dredze, I analyzed the manifestation of protests and riots on Twitter in an in-depth analysis of two distinct large-scale events to… Show more I successfully fine-tuned an LLM (specifically BART) to improve text summarization results for my capstone project under the supervision of Prof. Mark Dredze. My area of research was the summarization of Twitter tweets discussing specific topics (or hashtags) using an extractive/abstractive summarization approach. Previously under the supervision of Prof. Dredze, I analyzed the manifestation of protests and riots on Twitter in an in-depth analysis of two distinct large-scale events to help forecast civil unrest using tweets. The insights I was able to glean helped make decisions which in turn increased prediction accuracy and model interpretability. We presented this work at the Workshop on Noisy User-generated Text at EMNLP 2021. Show less I successfully fine-tuned an LLM (specifically BART) to improve text summarization results for my capstone project under the supervision of Prof. Mark Dredze. My area of research was the summarization of Twitter tweets discussing specific topics (or hashtags) using an extractive/abstractive summarization approach. Previously under the supervision of Prof. Dredze, I analyzed the manifestation of protests and riots on Twitter in an in-depth analysis of two distinct large-scale events to… Show more I successfully fine-tuned an LLM (specifically BART) to improve text summarization results for my capstone project under the supervision of Prof. Mark Dredze. My area of research was the summarization of Twitter tweets discussing specific topics (or hashtags) using an extractive/abstractive summarization approach. Previously under the supervision of Prof. Dredze, I analyzed the manifestation of protests and riots on Twitter in an in-depth analysis of two distinct large-scale events to help forecast civil unrest using tweets. The insights I was able to glean helped make decisions which in turn increased prediction accuracy and model interpretability. We presented this work at the Workshop on Noisy User-generated Text at EMNLP 2021. Show less

  • Tech Vedika
    • United States
    • IT Services and IT Consulting
    • 100 - 200 Employee
    • Machine Learning Engineer
      • Jan 2020 - Dec 2020

      Successfully fine-tuned a transformer based sentiment analysis model for Tech Vedika's e-commerce analytics platform which would parse product reviews and visualize the overall sentiments for the various product features detected (eg. screen, processor, battery life etc. for laptops) in the form of a radar chart. I also converted this tool into an end to end feature using Flask APIs and database integration. Successfully fine-tuned a transformer based sentiment analysis model for Tech Vedika's e-commerce analytics platform which would parse product reviews and visualize the overall sentiments for the various product features detected (eg. screen, processor, battery life etc. for laptops) in the form of a radar chart. I also converted this tool into an end to end feature using Flask APIs and database integration.

  • Western Digital
    • United States
    • Computer Hardware Manufacturing
    • 700 & Above Employee
    • Data Science Intern
      • May 2019 - Jul 2019

      I used historical data of in-house tool (eg. CAD, EDA etc.) usage to forecast the number of licenses required for the subsequent year. I preprocessed large unstructured datasets using libraries like pyspark and pandas and performed feature engineering to identify key predictors. I also compared the viability of 2 different chatbots, IBM Watson Discover and AWS (using lambda) for in-house use. I built APIs to query chatbot request from a MySQL database. I used historical data of in-house tool (eg. CAD, EDA etc.) usage to forecast the number of licenses required for the subsequent year. I preprocessed large unstructured datasets using libraries like pyspark and pandas and performed feature engineering to identify key predictors. I also compared the viability of 2 different chatbots, IBM Watson Discover and AWS (using lambda) for in-house use. I built APIs to query chatbot request from a MySQL database.

  • Vicara
    • India
    • Software Development
    • 1 - 100 Employee
    • Software Developer
      • Jan 2018 - Dec 2018

      I helped raise over $100,000 in crowdfunding as an early software developer for a novel human augmentation startup founded by VIT alum. I develop ed the software interface that enabled hand gestures to be transformed into customizable commands on a PC. The gesture controls were customizable across different applications and could be seamlessly integrated into any workflow. I also worked with application APIs to provide deep integration with the gesture controls beyond simply simulating keyboard… Show more I helped raise over $100,000 in crowdfunding as an early software developer for a novel human augmentation startup founded by VIT alum. I develop ed the software interface that enabled hand gestures to be transformed into customizable commands on a PC. The gesture controls were customizable across different applications and could be seamlessly integrated into any workflow. I also worked with application APIs to provide deep integration with the gesture controls beyond simply simulating keyboard shortcuts. Show less I helped raise over $100,000 in crowdfunding as an early software developer for a novel human augmentation startup founded by VIT alum. I develop ed the software interface that enabled hand gestures to be transformed into customizable commands on a PC. The gesture controls were customizable across different applications and could be seamlessly integrated into any workflow. I also worked with application APIs to provide deep integration with the gesture controls beyond simply simulating keyboard… Show more I helped raise over $100,000 in crowdfunding as an early software developer for a novel human augmentation startup founded by VIT alum. I develop ed the software interface that enabled hand gestures to be transformed into customizable commands on a PC. The gesture controls were customizable across different applications and could be seamlessly integrated into any workflow. I also worked with application APIs to provide deep integration with the gesture controls beyond simply simulating keyboard shortcuts. Show less

  • CoArtha Technosolutions Pvt. Ltd.
    • India
    • Information Technology & Services
    • 1 - 100 Employee
    • NLP Intern
      • May 2018 - Jul 2018

      Successfully developed an NLP pipeline to 1. parse online healthcare job listing paragraphs using web automation tools such as Beautifulsoup and Selenium and 2. extract template details like required skills, responsibilities, job location etc. from the job descriptions to analyze industry requirements. For the information extraction task, I used SpaCy to build an NLP model by training a custom Named Entity Recognizer instead of using the default pretrained one. To train the NER, I had to build… Show more Successfully developed an NLP pipeline to 1. parse online healthcare job listing paragraphs using web automation tools such as Beautifulsoup and Selenium and 2. extract template details like required skills, responsibilities, job location etc. from the job descriptions to analyze industry requirements. For the information extraction task, I used SpaCy to build an NLP model by training a custom Named Entity Recognizer instead of using the default pretrained one. To train the NER, I had to build a dataset from scraped job listings and manually annotate the data. Show less Successfully developed an NLP pipeline to 1. parse online healthcare job listing paragraphs using web automation tools such as Beautifulsoup and Selenium and 2. extract template details like required skills, responsibilities, job location etc. from the job descriptions to analyze industry requirements. For the information extraction task, I used SpaCy to build an NLP model by training a custom Named Entity Recognizer instead of using the default pretrained one. To train the NER, I had to build… Show more Successfully developed an NLP pipeline to 1. parse online healthcare job listing paragraphs using web automation tools such as Beautifulsoup and Selenium and 2. extract template details like required skills, responsibilities, job location etc. from the job descriptions to analyze industry requirements. For the information extraction task, I used SpaCy to build an NLP model by training a custom Named Entity Recognizer instead of using the default pretrained one. To train the NER, I had to build a dataset from scraped job listings and manually annotate the data. Show less

  • SHEILD INDIA
    • India
    • Android Developer
      • Sep 2017 - Nov 2017

      Successfully developed a full stack android application using a firebase backend for in-house workflow management across functional divisions in the organization. Successfully developed a full stack android application using a firebase backend for in-house workflow management across functional divisions in the organization.

Education

  • The Johns Hopkins University
    Master of Science, Data Science
    2021 - 2022
  • Vellore Institute of Technology
    Bachelor of Technology, Computer Science
    2016 - 2020

Community

You need to have a working account to view this content. Click here to join now