Nishit Asnani

Nishit Asnani

About

Detail

Co-founder
California, United States

Timeline

work
Job
auto_stories
Publication

Résumé

Jobs verified_user 0% verified
  • Sybill
    Co-founder, Growth
    Sybill
    Sep 2020 - Current (5 years 10 months)
    Sybill helps sellers become superhumans. It tracks their action items across deals, prioritizes them, and executes on their behalf! Say goodbye to writing follow-up emails, internal comms, or manually customizing business cases - Sybill can detect what's needed and do it for you. Check out Sybill at https://www.sybill.ai
  • Stanford University
    Teaching Assistant
    Stanford University
    Apr 2020 - Jun 2020 (3 months)
    Helping run the course CS224U: Natural Language Understanding (cs224u.stanford.edu).
  • Moveworks
    Product Manager
    Moveworks
    Sep 2019 - Dec 2019 (4 months)
    Led the development of a customer facing dashboard by leveraging insights from enterprise customers, product, engineering and customer success teams.
  • Moveworks
    Machine Learning Engineer
    Moveworks
    Jun 2019 - Sep 2019 (4 months)
    Built a deep learning system for named entity recognition and entity discovery with a large unlabeled data set, using weak supervision to noisily label it autonomously. This helped automate part of the NLP stack.
  • W
    Co-Founder
    Wordsmith
    Oct 2018 - Jun 2019 (9 months)
    Built an online community of creative writers, who provide feedback to each other on their work in public and private groups. Ideated, designed and built the product. Conducted customer interviews, pitched to investors, pivoted twice, managed a team of three technical interns. Incubated in d.school Launchpad (https://www.launchpad.stanford.edu/), the most successful Stanford accelerator.
  • Stanford Artificial Intelligence Laboratory SAIL
    Graduate Student Researcher
    Stanford Artificial Intelligence Laboratory SAIL
    Sep 2018 - Mar 2019 (7 months)
    Designed and implemented a deep learning model to detect Tuberculosis in HIV infected patients using their chest X-rays and other symptomatic covariates. We demonstrate that physicians equipped with our model do better than those that diagnose the same set of patients without assistance from it. Our work was covered by NPR and has been published in Nature Digital Medicine.
  • I
    Instructor
    IIT Kanpur Computer Science Department
    May 2018 - Jun 2018 (2 months)
    Co-taught a crash course on machine learning at Indian Institute of Technology Kanpur (IITK) as part of a summer program organized by Association of Computing Activities, IITK. Designed the course structure, created the content, delivered the lectures and mentored projects.
  • S
    Software Engineering Intern
    May 2017 - Jul 2017 (3 months)
    Worked at the Mobile Device Management team to facilitate device enrollment for Microsoft devices using Google MDM.
  • Strideai Inc
    Machine Learning / NLP Intern
    Strideai Inc
    Dec 2016
    Helped build a platform for abstractive text summarization and sentence similarity using an encoder-decoder sequence to sequence model in TensorFlow.
  • N
    Machine Learning Intern
    New York Office IIT Kanpur
    May 2016 - Dec 2016 (8 months)
    Worked on generating document embeddings for documents to be used for tagging, using a Deep Belief Network. Used Apache Kafka for sending messages in an online fashion to the model and stored the responses in a Postgre SQL database
Education verified_user 0% verified
  • Stanford University
    Master of Science - MS, Computer Science
    Stanford University
    AI research before LLMs made AI mainstream. Worked in Prof Andrew Ng's lab to detect TB from chest X-rays using large multi-modal models. Applied AI research in text tonality, psychology, creative writing, and retail. Before applying AI to B2B sales.
  • Indian Institute of Technology Kanpur
    Bachelor of Technology (B.Tech, Computer Science
    Indian Institute of Technology Kanpur
    Fell in love with AI and English Literature. Figured out I wanted to build AI that could understand and generate language.
Projects (professional or personal) verified_user 0% verified
  • P
    Paraphrase Generation using Deep Generative Models
    Developed a model to generate multiple paraphrases for a source sentence using a variational autoencoder with control variables in the latent space used to specify the length of the paraphrase required. Achieved better BLEU scores than vanilla seq2seq models.
  • C
    COMBINATORIAL GAME THEORY
    1. Studied the basic concepts of combinatorial game theory and the analysis of various games like Nim, Hex and Tic Tac Toe as a Discrete Mathematics Course Project. 2. Studied the mathematical treatment of games and some of the results that can be derived from it.
  • E
    Explicit Exploration in Random Forests
    Objective was to introduce an explicit exploration control knob in random forests for multi-label learning Showed empirically that using UCB / Thompson Sampling over the features at each level of a random forest for splitting each node consistently beat other criteria like empirical mean (with ranking loss) for major binary classification datasets (spam, MNIST etc)
  • D
    Disambiguating sentiment: An ensemble of humour, sarcasm, and hate speech features for sentiment classification
    Humour, sarcasm and hate speech detection in text using ensemble CNN models.
  • C
    CheXaid: deep learning assistance for physician diagnosis of tuberculosis using chest x-rays in patients with HIV
    Tuberculosis (TB) is the leading cause of preventable death in HIV-positive patients, and yet often remains undiagnosed and untreated. Chest x-ray is often used to assist in diagnosis, yet this presents additional challenges due to atypical radiographic presentation and radiologist shortages in regions where co-infection is most common. We developed a deep learning algorithm to diagnose TB using clinical information and chest x-ray images from 677 HIV-positive patients with suspected TB from two hospitals in South Africa. We then sought to determine whether the algorithm could assist clinicians in the diagnosis of TB in HIV-positive patients as a web-based diagnostic assistant. Use of the algorithm resulted in a modest but statistically sig
  • W
    Wordsmith
    Wordsmith was a platform for creative writers to share their work and receive feedback from AI on their tone, plot cohesion and consistency, and more. It also created a network of aspirational writers sharing feedback and promoting each other's work.
  • L
    LEARNING ATARI GAME STRATEGIES USING DEEP REINFORCEMENT LEARNING
    1. Created an AI agent which learnt to play a number of Atari Games and defeated Atari hard-coded AIs, using same set of hyper parameters. 2. Implemented Q Learning algorithm described in seminal papers by Google Deepmind and used a Convolutional Double Dueling Deep Neural Q-network to approximate the Q value function. 3. Used Bellman Equation and followed deep policy network based approach to implement Monte Carlo policy gradients.
Publications verified_user 0% verified
  • N
    CheXaid: deep learning assistance for physician diagnosis of tuberculosis using chest x-rays in patients with HIV
    NPJ Digital Medicine Nature Partner Journals
    Aug 2020
    Tuberculosis (TB) is the leading cause of preventable death in HIV-positive patients, and yet often remains undiagnosed and untreated. Chest x-ray is often used to assist in diagnosis, yet this presents additional challenges due to atypical radiographic presentation and radiologist shortages in regions where co-infection is most common. We developed a deep learning algorithm to diagnose TB using clinical information and chest x-ray images from 677 HIV-positive patients with suspected TB from two hospitals in South Africa. We then sought to determine whether the algorithm could assist clinicians in the diagnosis of TB in HIV-positive patients as a web-based diagnostic assistant. Use of the algorithm resulted in a modest but statistically sig
  • W
    Disambiguating Sentiment: An Ensemble of Humour, Sarcasm, and Hate Speech Features for Sentiment Classification
    Workshop on Noisy Usergenerated Text WNUT Empirical Methods in Natural Language Processing EMNLP
    Sep 2019
    Due to the nature of online user reviews, sentiment analysis on such data requires a deep semantic understanding of the text. Many online reviews are sarcastic, humorous, or hateful. Signals from such language nuances may reinforce or completely alter the sentiment of a review as predicted by a machine learning model that attempts to detect sentiment alone. Thus, having a model that is explicitly aware of these features should help it perform better on reviews that are characterized by them. We propose a composite two-step model that extracts features pertaining to sarcasm, humour, hate speech, as well as sentiment, in the first step, feeding them in conjunction to inform sentiment classification in the second step. We show that this multi-
This is a community-created genome.