Machine Learning at LinkedIn
9 Years of Experience
San Francisco, California, USA
97400******
0
0
Not Available
Speaking Engagements: 1. Conducted a 6-days workshop on "Machine learning and Deep learning for NLP" at NIT Warangal during Dec 22 - Dec 27, 2017. Background: 1. Strong Programming Skills: C++, Java, Python, Numpy, Tensorflow. Experience working into Supervised learning and Unsupervised learning (intermediate). Currently exploring reinforcement learning. 2. I have experience working in core-NLP projects like Text classification, Dependency Parsing, Named Entity Recognition, Dialogue systems and chat-bots where we apply appropriate machine learning and deep learning based architectures to build the system. 3. I like working on NLP/ML/DL based projects which involves end-to-end role right from getting data to training/testing model to deploying it into production. 4. I like to read relevant research papers in NLP domain such as Dialogue systems and chat-bots, Question-answering, representation learning etc. I try to attend university offered courses in domain of NLP/ML/DL. 5. I have worked upon Voice Technology based communication System usually deployed in devices like Smart Phones etc. I successfully analysed performance of various Phonetic algorithms and proposed a comparative study over Foreign Language Support. Built Spanish Name-matching algorithm for deployable in Mobile devices. 6. I have been a part of Samsung C-Lab project handled by team of 4 members. Completed project successfully within given time frame. Presented the solution at HQ (South Korea)
LinkedIn, SaaS/Cloud Product, Computer Software
LinkedIn, LinkedIn, LinkedIn, UC Santa Cruz Natural Language Processing Master's Program, TAIGER, IPsoft, Samsung Research India, Bangalore, National Institute of Technology Warangal, Samsung Research Institute - Bangalore, Indian Institute of Technology, Bombay
Job Title : Sr. Applied Research Engineer - NLP, ML, DL
Company name : LinkedIn
Period : August 2021 - Present
Location : Mountain View, California, United States
Job Title : Sr. Applied Research Engineer - NLP, ML, DL
Company name : LinkedIn
Period : September 2020 - August 2021
Location : Bengaluru, Karnataka, India
Job Title : Applied Research - NLP,ML,DL
Company name : LinkedIn
Period : May 2018 - September 2020
Summary : 1. I am working on problems in the domain of Content Quality in LI Feed and Anti-Abuse use-cases from Spam and Low-quality content detection, Virality prediction, etc.
2. User activity modeling for Job Search suggestions in Extreme classification setting.
Location : Bengaluru Area, India
Job Title : Capstone Mentor to the NLP MS Program
Company name : UC Santa Cruz Natural Language Processing Master's Program
Period : March 2022 - Present
Summary : I am working as a Capstone Mentor for the NLP MS Program at UCSC.
We are working with a group of four NLP MS students on Reverse Dictionary task. The task is aimed towards finding the right word from a human-typed definition (or dictionary gloss) of a word. We are also working to extend this system from en->en to truly multilingual -> multilingual.
Ex: "A device to talk to someone" -> Telephone/Phone
Students:
https://www.linkedin.com/in/utkarshug/
https://www.linkedin.com/in/kartikaggarwal98/
https://www.linkedin.com/in/anuroopjohnabraham/
https://www.linkedin.com/in/archit-bose/
Location : San Francisco Bay Area
Job Title : NLP Engineer
Company name : TAIGER
Period : January 2018 - April 2018
Summary : I was working to build Chatbots for Singapore Government Clients like Housing & Development Board etc.
Location : Singapore
Job Title : Research And Development Engineer
Company name : IPsoft
Period : July 2016 - January 2018
Summary : 1. Multi-Turn Contextual Dialogue system: Implemented Multi-Turn Contextual Dialogue system with Episodic Memory, Topic and slot injection into CLSTM based deep learning model. Currently in Production.
2. Dependency Parsing: Implemented "A Fast and Accurate Dependency Parser using Neural Networks" in Tensorflow which is used for understanding sentence’s Dependency tree structure and works as an additional feature for tasks such as Intent detection etc.
3. Convolutional Neural Network for Sentence Classification.: Implemented "Convolutional Neural Networks for Sentence Classification" in Tensorflow to be used in text classification tasks. Currently in production.
4. Named Entity Recognition: Worked on NER using Bi-LSTM + char-CNN based model on OntoNotes dataset, analyzed performance with Google NLP API on real time QA data. Also explored Fine grained Entity recognition system (FIGER - 112 entities) which can help to improve on tasks such as Q/A, coref and relation extraction.
Location : Bengaluru Area, India
Job Title : Software Engineer
Company name : Samsung Research India, Bangalore
Period : July 2015 - June 2016
Summary : 1. Spell Correction: Built Spell correction system based on Google confusion matrix, word/char count statistics and probability based scoring. Observed performance gain when using Metaphone word-code rather than actual word. Submitted 2-page write-up for paper presentation at SRIB.
2. CODE: Computer Personal Assistant: Basic "Speech + Text" based system that can perform actions like launching apps on your system. Used Microsoft Speech API, NLTK tokenizer, POS tagger and CFG parser to parse the input utterance and map it to corresponding response.
3. EasyShop: Successfully implemented SRIB CLAB15 winner idea EasyShop. Presented the idea at Samsung HQ South Korea. Got positive feedback by HQ.
Location : Bengaluru Area, India
Job Title : Placement Co-ordinator - Tranining and Placement Section
Company name : National Institute of Technology Warangal
Period : 2014 - 2015
Summary : Led campus placements of various engineering stream (B.Tech, M.Tech, M.C.A., M.Sc.) and achieved significant attraction of Software Companies, Core companies and leading Teaching Institutions. We achieved best metrics in terms of number of successfully placed students as well as much improved avg. CTC for B.Tech CS and M.Tech CS curricula.
Location : Warangal Area, India
Job Title : Intern
Company name : Samsung Research Institute - Bangalore
Period : May 2014 - June 2014
Summary : 1. Phonetic Algorithm: Worked on Various Phonetic Algorithms and Analyzed their performance over various languages and Developed Android tool for Spanish Name Matching for Galaxy S6
Job Title : Summer Intern
Company name : Indian Institute of Technology, Bombay
Period : May 2013 - June 2013
Summary : 1. FOSSEE IIT Bombay and MHRD India: Programming of Statistics based Real world Mathematical problems (Probability, Hypothesis, Graphs and charts, Mathematical operations - Mean, Median, Regression etc.) into Python.
project link: http://tbc-python.fossee.in/book-details/134/
link: http://fossee.in/
Title : Financial Markets
Period : December 2021 - Present
Summary : coursera.org, https://coursera.org/share/21fde5d589bd7889dfea35b87c410507
Issuing Authority : Yale University
Title : Structuring Machine Learning Projects
Period : November 2017 - Present
Summary : QLUAZ5Q84LE4, coursera.org, https://www.coursera.org/account/accomplishments/verify/QLUAZ5Q84LE4
Issuing Authority : Coursera
Title : Neural Networks and Deep Learning
Period : October 2017 - Present
Summary : ZRZFQHXHF7NL, coursera.org, https://www.coursera.org/account/accomplishments/verify/ZRZFQHXHF7NL
Issuing Authority : Coursera
Title : Machine Learning - Stanford University
Period : September 2016 - Present
Summary : GN44TY7HPEQD, coursera.org, https://www.coursera.org/account/accomplishments/verify/GN44TY7HPEQD
Issuing Authority : Coursera Course Certificates
Title : FOSSEE - Python Companion Project (MHRD)
Period : April 2014 - Present
Summary : fossee.in, http://tbc-python.fossee.in/book-details/134/
Issuing Authority : Indian Institute of Technology, Bombay
Title : [Patent] Unified intent understanding for deep personalization
Publication time : 2021
Summary : In an example embodiment, user interactions with a graphical user interface are modeled to derive an efficient representation that is highly avai
English (Professional Working), Hindi (Native Or Bilingual)
Award : 6th Position in Samsung Hack2Innovate Challenge - SMS classification
Issuer : Samsung Research India - Bangalore
Date : 11 2017
Summary : We developed few Machine Learning and Deep Learning based approaches to classify SMS messages into 3 categories (Ham, Spam, Info). Since the data was very sparse in terms of vocabulary, we performed standard pre-processing and built various models (SVM, NB, CRF, word-CNN, char-CNN + word-CNN, Bi-LSTM) and tried ensemble approach as well. Best performance was achieved on char-CNN + word-CNN based architecture which can learn embeddings for out-of-vocab words as well. In order to prevent overfitting, we employed strategies like dropout, L2 weight decay along with explicit Gaussian noise in input layer.We secured 6th position among 20 teams with an weighted F1 score of 95.48 while top team achieved 95.90 F1 score.
Show More