FlowCV Logo
Profile

Experienced AI educator and researcher with over 3.5 years of academic teaching and 1+ years of hands-on industry experience in NLP, computer vision, and LLMs. Proven ability to design learner-centric technical content, mentor diverse learners, and contribute to open-source AI initiatives. Deeply passionate about inclusive AI education, ethics in machine learning, and creating world-class curricula that inspire and empower global learners.

Education

Pursuing PhD under the supervision of Prof. S. Verma and Dr. A. S. Raghuvanshi in the field of Deep Learning X Cardiac MRI.

Research Scholar and Teaching Assistant, EEE, Signal Processing (7.57 CPI)

  • •I was pursuing PhD, supervised by Prof. M. K. Bhuyan in collaboration with Iwahori Lab for my project.
  • •My work was on Facial Expression Recognition.
  • •Coursework: Machine Learning, Linear Algebra and Optimization, Advanced Topics in Random Process, Probability and Stochastic Processes.
  • •MOOCS: CS 229 (Machine Learning), CS 231n (Deep Learning for Computer Vision), CS 224W (ML with Graphs).
  • •Teaching: Advanced Random Processes, Basic Electronics, EE 524 Introduction to ML Lab , EE 521 Digital Signal Processing Lab (I taught for a span of 6 semesters)
  • •Taught and supported undergraduate and postgraduate students in the mentioned subjects.
  • •Designed and conducted tutorial sessions, labs, and hands-on workshops for classes of 40–100 students.
  • •Contributed to curriculum planning, assignment design, and grading rubrics.
  • •Provided one-on-one mentoring to students and supported their academic development.

    • Masters, ECE, Robotics (8.42 CPI)

  • •Coursework: Mathematics, Linear Algebra, Digital Image Processing, Robotics, Mechatronics, Artficial Intelligence, Machine Learning.
  • •Volunteered in DRDO Robotics Unmanned Systems Exposition (DRUSE) held at DIAT Pune
  • •Worked as a project intern at CAIR, DRDO for final year project.
  • •Honorable Mention, AlphaPilot Lockheed Martin AI Drone Racing Innovation Challenge (Team Hawkeye)

    • Bachelors, ET&TC (7.97 CPI)

  • •Coursework: Core subjects from Electronics, Telecommunication and Electrical Engineering. Digital Image Processing, Pattern Recognition and Artificial Neural Networks as elective courses.
  • •Core Member and volunteer, Raaga (The music club of NIT Raipur), Tabla player in the annual musical fest Shruti.
  • •Internship from Doordarshan Kendra Raipur and TPIPRD Nimora.
    • Professional Experience
    05/2023 – 07/2023 | Remote, India
  • •My work involved Research and Development on LLMs, Transformers, LangChain for providing solutions to company usecases.
  • •Research and Development on Natural Language Querying, text2sql models and datasets like Spider, WikiSQL etc. Text to SQL can be used to replace writing long queries and get insights from data faster.
  • •Inference on Transformer models, dataset preparation for fine-tuning transformer models for the same, experience with HuggingFace models and LangChain for agents.
    • 08/2022 – 09/2022 | Remote, Japan
  • •My role involved working with a diverse team, developing POCs, research and development in the domain of AI-assisted dashcams for safer mobility.
  • •Worked on ANPR, Traffic Sign Recognition, object detection for Japan using Computer Vision.

    • Defense Laboratory working on Robotics and AI

  • •Worked on GPS Location Estimation and Object Detection modules of the drone.
  • •Developed a mathematical model for estimating depth of object using GPS and image information. Recorded ROS bag video files were used for this.
  • •Packages used: csi_cam, mavros, darknet_ros. ROS Version: Kinetic. Operating System: Ubuntu16.04 LTS.
    • Skills
    Python
    Keras
    PyTorch Geometric
    Transformers
    Scikit-learn
    PyTorch
    Spektral
    LangChain
    Open Source
  • •Skills: Course curation and development, course outline creation, drafting issues, GitHub collaboration, and pushing content via PRs with multiple reviews and iterations from unit members and HF Team. PRs merged till now.
    • 08/2023 – 11/2023 | Remote, International
  • •Selected as a mentee, for the 3rd Cohort of KaggleX BIPOC Mentorship program, which aims to increase the representation, career growth for BIPOC.
  • •This program pairs early career data scientists/Kagglers with experienced mentors for career related discussions and learning opportunities.
  • •One of the few mentees to receive a career development grant of $500 to support my career journey and learning.
  • •Final Project: Exploring BIPOC Representation in Data Science, PPT | Recording | Certificate

    • The aim of the project was to explore the representation of BIPOC in data science with factors like gender-ratio, unemployment, race and ethnicity.

    Bhashini Language Contributor, MEITY, Government of India
    09/2023
  • •Contributing to BhashaDaan Beta under NLTM (National Language Translation Mission) under Likho India track.
  • •Made around150+ contributions combinedly in translate task and validation task.
  • •The aim of this program is to crowdsource language datasets to enrich Indian languages because many of them are low resource, into digital products and services that can be used all over India by citizens in their own languages.
    • 07/2023 – 10/2023 | Remote, International
  • •Selected as a mentee (Padawan) for OpenMined's R5Q3 Padawan Mentorship Program.
  • •I was paired with Rasswanth from OpenMined, my mentor for the program and the program offered independent learning and support on OpenMined's tech stack like PySyft.
  • •Started off with a very small contribution to the PySyft Repo: Issue #7701.
  • •Also got the opportunity to explore the OpenMined and interesting domains like Policy AI. Certificate of Completion
    • Kaggle Datasets Expert, Kaggle
    05/2023

    Became Dataset Expert on Kaggle (1X Kaggle Expert). Best Rank: 555 out of 96899. Check my Kaggle Profile.

    The aim of this community sprint is to build interactive demos from the scikit-learn documentation and afterward, contribute the demos directly to the docs. Some contributions can be found on this HF Space.

  • •Participated and won the 3-week long community sprint, on Gradio themes.
  • •The sprint involved 5 special tracks, out of which I participated in the 2nd track (sector-specific theme).
  • •I proposed a custom theme for Scikit-learn which was the winner for 2nd Track. I won $50 worth of goodies from HuggingFace Store and a pro-membership for my HF Account.
    • Projects
    Graph Based Facial Expression Recognition
    07/2019 – 01/2023

    Ph.D. Work

  • •Facial Expression Recognition on different datasets like: JAFFE, CK+, FER-2013, AFEW, SFEW, AffectNet, CELC.
  • •Tasks: License permissions for datasets, data cleaning, EDA, data preprocessing, frame extraction, landmark extraction (68 points), metadata extraction, reproducing baselines, model training, validation, and inference.
  • •Models and frameworks: CNN (Keras), Graph Neural Networks (Spektral, PyTorch Geometric).
  • •Participated in various FER competitions, got one publication from the Indian Contextual Emotion Learning Challenge hosted by IEEE Automatic Face Gestures Conference, 2021.
  • •Future work: Multimodal emotion recognition, temporal graph based FER, FER applications to self-driving cars.

    • Open Source Project, GitHub and Kaggle

  • •Used the Construction Site Safety Dataset by Roboflow, to detect PPE (hardhats, safety vests, masks, machinery). The dataset is made public on Kaggle for community usage.
  • •Wrote a Kaggle notebook with an extensive EDA, and pre-trained a custom object detection model pre-trained on COCO128 dataset.
  • •YoloV8 Nano model was trained for 100 epochs on the custom dataset and was evaluated on test videos and images.
  • •Frameworks used: Pandas, Ultralytics YOLOV8

    • Open Source Project, GitHub

  • •Built, trained, validated and evaluated a vanilla CNN model for traffic sign detection task on German Traffic Sign Recognition Benchmark dataset.
  • •Extensive EDA with visualizations, saved trained model for inference. The classifier predicts traffic signs with an accuracy of 97.9%.
  • •Frameworks used: Pandas, Keras

    • Coursera Capstone Project, Getting Started with TensorFlow 2

  • •Built, trained, validated and evaluated 2 models (MLP and CNN) for the image classification task on Street View House Number (SVHN) dataset.
  • •Preprocessed the data, visualized the data, implemented model checkpoint and early stopping callbacks. Obtained test accuracy of 78% using MLP and 89.77% using CNN.
  • •Frameworks used: Tensorflow 2, Keras

    • Virtual Presence Experience for Graduation Students

  • •Organized and conducted an interactive Vitual Presence experience for 1800 graduating students. The project was developed in 10 days.
  • •The program was divided in 4 slots with 500 students capacity and a password was given for entry.
  • •The effort was recognized widely by local and national media. IIT Guwahati became one of the institutes after IIT Bombay to conduct a virtual convocation.
  • •Tools used for design: Autocad, Gather town.
    • Certificates
    Getting Started with TensorFlow 2

    (05/2022), Coursera

    Course offerred by Imperial College of London

    Data Visualization with Plotly

    (03/2022)

    Coursera Project Network

    Emotion AI: Facial Key-points Detection

    (10/2020)

    Coursera Project Network

    Web Scraping with Python + BeautifulSoup

    (10/2020)

    Coursera Project Network

    Mathematics for Machine Learning: Linear Algebra

    (04/2022), Coursera

    Course offerred by Imperial College of London

    CVIT 5th Summer School on Artificial Intelligence, IIITH

    (08/2021 - 09/2021)

    Summer School organized by IIIT Hyderabad

    Programming for Everybody (Getting Started with Python)

    (01/2019), Coursera

    Offerred by University of Michigan

    Publications

    This paper is the outcome of the Contextual Emotion Learning Challenge organized by IEEE FG 2021.

    This paper is the outcome of my final year Masters Project where I worked with CAIR, DRDO (Defence Lab of India).