Akshat Joshi

AI/ML Engineer at ClusterDev
  • Claim this Profile
Contact Information
us****@****om
(386) 825-5501
Location
New Delhi, Delhi, India, IN

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

  • Huggingface DL course part 1
    Hugging Face
    Jan, 2022
    - Nov, 2024
  • Managing Machine Learning Projects
    Coursera
    Jan, 2022
    - Nov, 2024
  • NLP course by HSE
    Coursera
    Jan, 2021
    - Nov, 2024
  • Deep Learning Specialization
    Coursera
    Jan, 2020
    - Nov, 2024
  • Intro to Machine Learning
    myTectra
    Sep, 2018
    - Nov, 2024

Experience

  • ClusterDev
    • India
    • Technology, Information and Internet
    • 1 - 100 Employee
    • AI/ML Engineer
      • Jan 2021 - Present

      • Led AI initiatives, building features for 10M+ daily users, across ASR, TTS, NLP modules in languages like Hindi, Bengali, Telugu, Tamil etc. • Developed and managed a conversational AI product in the voice-telephony domain using open-source Asterisk PBX, Rasa, multilingual PLMs and our in-house technologies. • Optimized data pipelines, reducing dataset preparation from 6 days to a few hours and dev efforts considerably. • Deployed A/B tested models on fault-tolerant EC2 servers, serving hundreds of thousands of users. • Achieved two 2nd places in IITM ASR challenge against top AI teams like JHU, yellow.ai, Uniphore and IITs. Show less

  • Voicezen India (acquired by Sprinklr)
    • Gurugram, Haryana, India
    • AI/ML Engineer
      • Jul 2020 - Dec 2020

      • Worked on conversational AI projects, such as ASR, TTS, NLP; building Sequence2Sequence models with OpenSeq2Seq and DeepSpeech. • Designed and managed regex and bot flows for voice bots and developed text-to-speech solutions using Non-Autoregressive Transformer-based models. • Orchestrated workflows for multiple voice artists, handling scheduling, timelines, and raw audio processing. • Started learning Kaldi to build hybrid DNN-HMM based ASR systems. • Worked on conversational AI projects, such as ASR, TTS, NLP; building Sequence2Sequence models with OpenSeq2Seq and DeepSpeech. • Designed and managed regex and bot flows for voice bots and developed text-to-speech solutions using Non-Autoregressive Transformer-based models. • Orchestrated workflows for multiple voice artists, handling scheduling, timelines, and raw audio processing. • Started learning Kaldi to build hybrid DNN-HMM based ASR systems.

  • Sprinklr
    • United States
    • Software Development
    • 700 & Above Employee
    • Machine Learning
      • Feb 2020 - Jun 2020

      • Addressed sentiment analysis challenges by building, training, and tuning large deep learning models on extensive text datasets using Tensorflow. • Engaged in research, reading cutting-edge papers and completing a deep learning specialization to enhance skillset and approach. • Customized DeepMoji, a BiLSTM, to deliver innovative solutions for text classification. Note: Full-time position was planned for April, but circumstances changed due to the COVID-19 pandemic. • Addressed sentiment analysis challenges by building, training, and tuning large deep learning models on extensive text datasets using Tensorflow. • Engaged in research, reading cutting-edge papers and completing a deep learning specialization to enhance skillset and approach. • Customized DeepMoji, a BiLSTM, to deliver innovative solutions for text classification. Note: Full-time position was planned for April, but circumstances changed due to the COVID-19 pandemic.

  • Razorpod
    • India
    • Advertising Services
    • 1 - 100 Employee
    • Software Devoloper
      • Sep 2019 - Feb 2020

      • Finetuned an open-source LLM for text generation, providing innovative solutions for the copywriting team. • Developed a robust Web Crawler using Selenium, BeautifulSoup, and Regex to gather essential data. • Conducted daily activities in bash scripting, python scripting, data scraping and wrangling. • Created a Web Service running as a cronjob, designed to collect RSS feeds, optimizing information collection. • Finetuned an open-source LLM for text generation, providing innovative solutions for the copywriting team. • Developed a robust Web Crawler using Selenium, BeautifulSoup, and Regex to gather essential data. • Conducted daily activities in bash scripting, python scripting, data scraping and wrangling. • Created a Web Service running as a cronjob, designed to collect RSS feeds, optimizing information collection.

  • Graphnile
    • United Kingdom
    • Research Services
    • 1 - 100 Employee
    • Data Science Intern (Trainee)
      • Oct 2018 - Oct 2019

      • Scraped and wrangled large datasets, employing effective techniques for diverse applications. • Conducted image classification on various datasets including geographical and medical imagery. • Developed a Python based topic modeller using Latent Dirichlet Allocation to extract top topic candidates from articles. • Collaborated on Data Science for Social Good projects, researching and reviewing literature for state government initiatives in traffic management and energy conservation in Karnataka. Show less

  • myTectra
    • India
    • Education Administration Programs
    • 1 - 100 Employee
    • Machine Learning Course
      • Jul 2018 - Sep 2018

      Learned theory of various Machine Learning models like SVM, XGBoost and got introduced to Deep learning through Keras. Learned theory of various Machine Learning models like SVM, XGBoost and got introduced to Deep learning through Keras.

  • AmbitionBox
    • India
    • Technology, Information and Internet
    • 1 - 100 Employee
    • Machine Learning Intern
      • Feb 2018 - Jun 2018

      Automating Processes Using Machine Learning and Natural Language Processing Techniques • Developed a Company Name Tagger using multi-text classification for record linkage, enhancing the system's efficiency. • Built a Company Reviews-Gibberish Classifier to differentiate between meaningful reviews and irrelevant text. • Created a Python based extractive text summarizer that scores sentences and words based on term frequency and produces top candidates as a summary. Automating Processes Using Machine Learning and Natural Language Processing Techniques • Developed a Company Name Tagger using multi-text classification for record linkage, enhancing the system's efficiency. • Built a Company Reviews-Gibberish Classifier to differentiate between meaningful reviews and irrelevant text. • Created a Python based extractive text summarizer that scores sentences and words based on term frequency and produces top candidates as a summary.

  • Naukri.com
    • Technology, Information and Internet
    • 700 & Above Employee
    • Software Developer Intern
      • May 2017 - Jun 2017

      Automated Content Extraction from Webpages and Removal of Clutter • Created a Python application that extracts the main content from webpages and removes clutter like footers, pop-up ads, and extraneous links. • Utilized libraries like BeautifulSoup4 and Urllib2 for HTML parsing, employing regex for pattern matching. • Implemented an open-source algorithm using heuristics to identify and score positive and negative portions of the webpage, enhancing user readability. Automated Content Extraction from Webpages and Removal of Clutter • Created a Python application that extracts the main content from webpages and removes clutter like footers, pop-up ads, and extraneous links. • Utilized libraries like BeautifulSoup4 and Urllib2 for HTML parsing, employing regex for pattern matching. • Implemented an open-source algorithm using heuristics to identify and score positive and negative portions of the webpage, enhancing user readability.

Education

  • DIT UNIVERSITY
    Bachelor of Technology - BTech
    2014 - 2018

Community

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