L o a d i n g
Data Scientist & AI/ML Developer

Hello, I’m
Kinshuk Gaurav

I've done remote work for companies, consulted for pre-startups, mentored juniors, and collaborated with talented people to create advance technologies and methods for both business and consumer use.

2

Years of
Experiences
Working Domain

I Am Currently Working In...

8 Projects
Data Science

Data Analytics, Business Analytics, Business Development Analysis, Data Engineering, Data Visualizations, Data Pre-processing.

8 Projects
Machine Learning

Linear Regression, Random Forest, Recommendation System, Decision Trees, Logistic Regression, Supervised Learning, Un-Supervised Learning, SVM.

8 Projects
Computer Vision

Image Processing, Open-CV, Segmentation, Detection, Tracking.

8 Projects
Deep Learning

Neural Networks, Deep Neural Networks, LSTM, GRU, RNN, CNN, U-Net, GAN's, AutoEncoder.

8 Projects
Artificial Intelligence

NLP, LLM, Artificial Intelligence, ANN, Transformers, BERT, LSTM, Agentic AI, Ai agents, Retrival Augmented Riality (RAG), Graphify, SDD (Specc Driven Development), Prompt Engineering.

8 Projects
MLOps

Kafka, AWS, StreamLit, Github, Azure.

About Me

I'm Developer and Researcher fresher working to advance AI capabilities.

I enjoy turning complex problems into simple, beautiful, and intuitive designs, using my skills to create effective AI solutions.

Product Design
Product Development
AI Development
Research
Email Me
kgaurav11286@gmail.com
My Skills

My Advantages

Data Science
95%
VLM's
75%
Data Visualizations
95%
Image Processing
90%
Tensorflow
75%
Python
80%
Agentic AI & RAG
80%
My Resume

Experience

Feb 2025 - Present
Junior Data Scientist
Dash technologies

  • Designed and shipped a real-time Borescope obstruction detection pipeline: curated and annotated proprietary dataset, trained a YOLOv8-based model achieving 90% mAP at 15 FPS on edge hardware, prototyped Raspberry Pi deployment reducing on-site inspection time by ~60%
  • Engineered a production RAG pipeline using LangChain and vector embeddings on a corpus of 500+ enterprise documents, achieving ~90% retrieval accuracy and reducing average LLM response latency by ~40% (from ~8.3s to ~4.8s) through query optimisation and semantic chunking strategies.
  • Built and deployed an agentic AI document processing system using LangGraph — orchestrating multi-step extraction, classification, and validation across heterogeneous document types, eliminated ~75% of manual processing effort for 3 business teams in production.

Sep 2024 - Feb 2025
Data Science - Intern
Dash technologies

  • Developed end-to-end pipeline for the document extraction using classification, direction, LLMs, and VLMs based model with achieving the 80% faster performance with optimized code and trained own model for better local performance due to custom data.
  • Deployed Power BI dashboards for the chemical information and sales analysis.
  • Suggested advanced solutions to achieve 50% quicker tasks using the AI, and learned GANs through Coursera Certification, helped me to perform tasks efficiently and innovatively using advanced generative models, like in creation of datasets, and code development.

Aug 2023 - April 2024
AI Researcher
Koita Centre for Digital Health (KCDH),IITB

  • Participated in diverse medical AI projects, including medical imaging and EEG signal analysis. Authored publications and contributed to peer-review processes for other researchers’ work.
  • Developed a novel loss function for enhancing medical image segmentation accuracy by 6-8%, leading to a publication at A-tier conference - ISBI24.

June 2023 - April 2024
Research Intern
Visual learning and Intelligence Lab, IITH

  • Developed a federated learning system for brain tumor segmentation, increasing training efficiency by 40% and model generalization by 30%.
  • Wrote TensorFlow code from scratch, improving performance by 10%, maintainability, and reducing bugs by 70%.
  • Created a new loss function and framework, cutting security vulnerabilities by 50%.

September 2022 - April 2024
Teaching Assistant
Pandit Deendayal Energy University

  • Revamped and deployed a solar power forecasting model, cutting manual training by 70% and achieving 97% accuracy with a 25% performance boost.
  • Conducted code reviews, mentored juniors, and supported research, improving project efficiency by 50% and boosting code productivity by 20%.

August 2021 - October 2021
Machine Learning Internship
Feynn Labs

  • Led a project with fellow interns, enhancing the team’s market dominance by 15% and boosting the client's tea business by 15% through integrated supply chains.
  • Advised on tea market segmentation in South India, analyzing data and external factors to achieve a 15% profit increase.

August 2021 - October 2021
Summer Research Internship
Shiraz University of Medical Science

  • Led a team to analyze sleep disorder data, improving identification accuracy by 10% through uncovering hidden patterns and trends.
  • Published research findings in a reputable journal, aiding doctors in identifying early sleep disorders in 10% more cases.

Education

2022 - 2024
M.Tech - Data Science
Pandit Deendayal Energy University

Relevant Coursework: Computer Vision, Deep Neural Networks, Machine Learning, Time Series Data Analysis, Pattern Recognition and Machine Learning, Explainable AI, NLP, Cloud Infrastructure for AI.

2018 - 2022
B.E. - Computer Science and Engineering
Gujarat Technological University

Relevant Coursework: Programming languages (C/C++, Java, Python), Data structures and algorithm, Mathematics (Statistics, Calculus, Discrete math, Algebra), DBMS, IoT, Artificial Intelligence.

2017 - 2018
XII
Kendriya Vidyalaya No.2 E.M.E.

Relevant Coursework: Science (Physics, Chemistry, Mathematics), English, Computer Science (C++), Physical Education.

My Portfolio

Developing Advance Solution
To Scale The Product Design.

My Publications

Publications of My Awesome Research Works

I get involved in research work as part of collaborations, my faculty's work, and my keen interest to solve problems and provide more valuable solution.

Resent (Under Review) - AAAI
  • Developed medical image segmentation with data heterogeneity over privacy concerns.
  • Introduces a scale-adaptive federated framework with dynamic weighted aggregation.
  • Developed novel loss function to improve spatial accuracy and feature similarity.
  • Uses an epistemic uncertainty layer to enhance model robustness and handle data variability.
Best Practical Paper - AI x PAC workshop (Under AI xIA)
  • Used thermal infrared images to detect faces in low-light conditions.
  • Trained deep learning model for enhances detection accuracy.
  • The new method outperforms traditional face detection techniques in low-light scenarios.
  • The approach is effective for various poses and expressions, with potential uses in surveillance and security.
Go to Other Publications

Total number of Work Accomplished by me till now

12

Total Projects

5

Total Publications

2

Awards / Achievements

3

Collaborative Works (Research/Project)
Details
Contact Me

Let's Get in Touch!