Subhankar Ghosh

AI Researcher

New York, New York, United States10 yrs 1 mo experience

Most Likely To SwitchHighly Stable

Key Highlights

Expert in machine learning and deep learning algorithms.
Proven track record in statistical NLP and data analysis.
Strong software engineering background applied to real-world problems.

Stackforce AI infers this person is a Machine Learning and NLP specialist with a strong software engineering foundation.

Contact

subhankar2321@gmail.com LinkedIn

Skills

Core Skills

Machine LearningDeep LearningNatural Language ProcessingSoftware EngineeringSoftware Development

Other Skills

.NET3-gram contextual language modelAgile MethodologiesBiLSTMCC#C++Data AnalysisData StructuresData VisualizationDatabasesGenerative SpeechHDAJavaScriptKeras

About

Passionate about the Mathematics behind machine learning and deep learning algorithms. Also my software engineering skills help me apply machine learning and deep learning to real life problems.

Experience

Nvidia

2 roles

Senior Applied Research Scientist

Jan 2022 – Present · 4 yrs 2 mos

Generative Speech, LLMs in NeMo team

Generative SpeechLLMsNeMoMachine LearningDeep Learning

Deep Learning Engineer

Jul 2019 – Present · 6 yrs 8 mos

Deep Learning

Google

Software Engineering Intern

May 2018 – Aug 2018 · 3 mos · Pittsburgh, Pennsylvania

[NDA Restricted] I worked on outlier detection by implementing state-of-the-art sequence deep learning models and image embeddings.

outlier detectionsequence deep learning modelsimage embeddingsMachine Learning

Cline center for advanced social research

Graduate Research Assistant

Jan 2018 – May 2019 · 1 yr 4 mos · Urbana-Champaign, Illinois Area

I am working as a Research Assistant in the Analytics and Data Management Team pursuing research in statistical NLP.
Used an ensemble of 3-gram contextual language model and a character level BiLSTM to correct OCR spelling errors (82\% accuracy) on NY Times articles (Tensorflow, Keras)
Working on topic modelling techniques like LDA, HDA to detect terrorism related events and how they are reported in news articles.
Detection and tracking events/topics over a period of time and recognizing patterns of how they vary in that time period. Using dynamic topic models and seeded topic models.
Worked on automated spelling correction methods using language models built from Wikipedia data and finding near-duplicates.

Statistical NLP3-gram contextual language modelBiLSTMTensorflowKerasNatural Language Processing+1

University of illinois at urbana-champaign

Graduate Research Assistant

Sep 2017 – Dec 2017 · 3 mos · Urbana-Champaign, Illinois Area

I worked under Prof Vetle Torvik on stylometry of research papers to determine style of writing of different authors. Another problem that I worked on is predicting the age, experience and ethnicity of the author based on the style of writing determined from his/her research papers.

stylometrypredicting author demographicsNatural Language Processing

Microsoft india r&d pvt. ltd

2 roles

Software Engineer

Jun 2015 – Jul 2017 · 2 yrs 1 mo · Hyderabad Area, India

Worked closely with business and IT partners during early SDLC phases to understand the business capabilities and goals for the program. Developed engineering artifacts such as Logical Data Model, System Conceptual model, Component Interaction Model etc. Created windows and web services based on the Microsoft platform.Understood the existing systems, while analyzing upstream and downstream system impacts.Adept at working cross-group with other teams, as our development projects are often coordinated with other teams around the world.

SDLCLogical Data ModelSystem Conceptual model.NETSoftware Engineering