top of page

Hey there! I'm 

Diksha Shrivastava

ML Developer & Researcher

  • GitHub
  • Twitter
  • LinkedIn

About Me

marissa-grootes-ZCfGjkbJsU0-unsplash_edi

I'm Diksha, a 3rd-year student at Bennett university, pursuing BTech CSE with specialisation in AI. 

I have been into machine learning for three years now. My love for ML developed by simply seeing just how much I can positively create an impact.

 

Most vividly, I remember a thought experiment where individuals with mental disorders could be put into scenarios created with AI which can help them deal and change accordingly, allowing for better understanding human brain and making models iteratively closer to it. This was back in 2021 when I was in my first semester :) 

I owe a lot to Alan Turing, like everyone else.

I explored ML more and more afterwards. Another highlight of my first year is when I got excited after a couple of papers to proof using MRI whether music therapy really has a positive impact on mental disorders and if I can generate music which can be helpful for those individuals.

That led me to work on my first research project "Analysis of Neural Correlates of Different Music Genres using Machine Learning", which got selected for Fechner Day 2022, Sweden where unfortunately I couldn't go and present.

Since then, I've participated in and won multiple hackathons, received KaggleX grant, have been a part of Google's Women Engineers initiative, started contributing to open-sourced ML, worked on multiple projects, all the while working towards creating some value while getting more and more fascinated by research engineering marvels.

  • GitHub
  • Twitter
  • LinkedIn

Ongoing Projects I'm Happy About

ChessGAN 

& the Turing Test

Chess

Applying GANs for the first time (to the best of my knowledge) on Chess, to generate machines which can play at human level by making the maximum number of right moves while losing the game 50% of the time.

JuliaMind

ML Framework in Julia

Bayesian Flow Networks

Implementation, Application & Testing

bfn.png

Deep Learning + Bayesian Inference. Trying to implement the Bayesian Flow Networks paper, work on an application, compare with other models and test a hypothesis mentioned in the paper and another that seems a logical derivation.

julia_edited.png

Julia is a beautiful language that I've recently been exploring, however the ecosystem and support is very much limited. This is an effort to build a Deep Learning Framework, basing off PaddlePaddle while leveraging Julia. It'll be hopefully open for contribution soon :)

Soundscapes

Analysis of Music on Individuals with Intellectual & Development Disorders

Image by Dolo Iglesias

As the title says, exploring how music differently affects individuals with and without intellectual and development disorders

Vision

Navigation Aid for Individuals with Vision Impairments

Image by Aarif Sheikh

Google Maps' live view feature, but available for India and for blind people, made from scratch. 

WildCue

Recognition of Cues of Distress in Animals

Image by Undine Tackmann

Detecting and Raising alarm whenever animal behaviour suggests animals' declining health, preventing abuse.

Image by Annie Spratt

Till now, I've been a part of...

ivy_edited.jpg
Volunteer Machine Learning Developer
July '23 - Present
kaggleX.png
Google KaggleX Mentee
November '22 - March '22
  • Contributing to unify the fragmented ML stack across frameworks. Responsibilities include:

  • Getting methods adhering to the Array API Standard

  •  Adding new methods to Ivy's front-ends

  • Creating Jupyter notebooks on popular open source repos (PyTorch Geometric etc.)

  • Implementing popular architectures in Ivy's models repo etc.

  • Worked on implementing a research paper

  • Explored methods for Symbolic Music Understanding and Generation, and Music Information Retrieval.

  • Individual Development Plan includes contribution to open-sourced ML projects, participation in Kaggle Competitions

  • Selected as one of the Top 152 mentees across 15+ countries and 20+ timezones, among various data scientists, ML engineers, researchers and students.

  • Received $1000 grant and $1000 GCP credits.

  • Selected as Top 250 from 30,000+ applicants, Received 100% Scholarship.

  • Project - DeepFake Recognition: Deep Neural Network to Classify Real and Fake Videos

  • Project: ML Framework in Julia

Google TalentSprint WE Scholar
March '22 - Present
Research Team Co-Head
August '22 - August '23
  • Study of History and Development in Theory of Mind AI, Flower Recognition, Titanic Survival Prediction, Iksha - Neurological Tool with AI and Virtual Reality, Career Dendrogram with Machine Learning.

  • MentiWell: Journal with Text-Sentiment Analysis, Social-Media Activity Sentiment Analysis, Monitoring of Mental State with IOT, Directional Survey Covering Wide Range of Mental Disorders, Face/Body Language/Mood Detection, Empathical ChatBot, Music Generated with AI for Specific Disorders.

Past Work Gallery

Some Achievements

Google KaggleX: Selected as one of the Top 152 mentees across 15+ countries and 20+ timezones, among various data
scientists, ML engineers, researchers and students. Received $1000 grant and $1000 GCP credits.
Samsung’s Solve for Tomorrow Hackathon: Ranked 1096 from 18,000+ applicants, working on MentiWell.
Hackaccino Hackathon: Won working on ’Prediction and Prevention of Self-Harm with Mental Well-Being Analysis’.
Smart India Hackathon: Cleared Internal Round working on ’Lack of Information about Academic Events on a single
platform’. Researched on Career Dendrogram. Developed events interface with Flutter and Firebase.
WE - Cohort 4 (Google and TalentSprint): Selected as Top 250 from 30,000+ applicants, Received 100% Scholarship.
Researcher (Bennett University): Research Abstract selected for Fechner Day 2022 at Lund University, Sweden by
International Society for Psychophysics.
Academic (Bennett University): Topped the Computational Thinking and Programming Course (with Python) among
500+ students. Awarded for extraordinary performance in Hackathon.

My Skill Set

Languages: Python(Expert), Java(Proficient), C++(Good), Julia, MATLAB, JavaScript, SQL, Dart
Frameworks: TensorFlow, PyTorch, JAX, PaddlePaddle, MindSpore, Scikit-Learn, Keras, OpenCV, MediaPipe, FastAI
Tools: Git, Kaggle, Weights&Biases, H2O, VertexAI, HuggingFace, Google Cloud AI, Docker, MySQL, NoSQL
Libraries: Detectron2, Keras-Core, KerasCV, KerasNLP, Librosa, Music21, Falcon-40B, MatPlotLib, Pandas, Plotly
Soft Skills: Design Thinking, Research, Creativity, Problem Solving, Leadership

  • GitHub
  • Twitter
  • LinkedIn

Like my work? Drop a msg! :)

bottom of page