Sayantan Sadhu

Email: sayantan.sadhu1410@gmail.com
Phone: 7003302025

Hi there, welcome to my site. Hope you have a good time !
Full Stack Development, LLM trainings and a lot of things in between !

Experience

Systems Engineer

Tata Consultancy Services, India (Kolkata)
Aug 2023 - Present

Worked as a full stack developer with Angular JS for the frontend along with HTML and CSS and Java, SpringBoot for the Backend. My work mainly included:

  • Developed Multiple APIs (5+) using Spring Boot for developing key functionalities. Used JPA and Spring Data Repositories for fetching data from database and Angular JS for the front-end customization
  • Written functions and procedures in PL/SQL that reduces processing time by 5% for each call.
  • Eliminated approximately 300 defects (bugs) for the project.
  • Engineered core functionalities using RESTful API for automated policy renewal and complete policy workflow in Java; increased policy renewal efficiency by 4 times and decreased manual intervention by 90%. Also written scripts for automation testing using Selenium for the same

AI Algorithm Developer

Aarish Technologies, Canada (Remote)
October 2022 - April 2023

Worked on optimisation of the training script for the paper "FaceNet: A Unified Embedding for Face Recognition and Clustering" by Google and trained it on the VGG Face2 dataset with 3.3 million images. My work mainly included:

  • Writing code for creating triplets that is positive, negative and anchor image from the dataset and pass it to the dataloader.
  • Optimized and trained Face Net's embedding model using triplet loss on the VGG Face2 dataset containing 3.3 million images. Also successfully reduced model size by 25% through stage two quantization.
  • Tuning some important parameters like learning rate, margin of triplet loss and batch size.
  • Writing code that would truncate the parameters of the pytorch model to 8 bits at the time of training and also post training.

AI/ML Consultant

Averyx Group, Dubai (Remote)
June 2021 - December 2021

I was responsible for working on a project called TechScreener that recommends the best stocks to a customer from a pool of around 10k stocks. I had developed the ranking system and worked on data pipelining, development of strategies and calculating return percentages using a library called backtesting.py . The Backend for the application was build with django and postgresql and for the frontend technology we had used HTML, CSS and vanilla JavaScript

Machine Learning Educator Intern

Acadboost Private Limited, India (Remote)
August 2020 - October 2020

My job at Acadboost was to create a full fledged 25 hours course on Machine Learning for poeple who are absolutely beginner in Programming The Course included tutorials on Python, Numpy, Pandas, Matplotlib, Seaborn, Sci-kit Learn, thoery and application of some ML algorithms like Linear Regression, Logistic Regression, Decision Tree and Random Forest, some feature engineering techniques like handling missing values, outliers and categorical variables and some introduction to deep learning.

Core Team Member

Google Developers Students Club, KGEC
March 2020 - Present

My duty at my college's Developer Club is to promote Machine Learning to all the students interested in ML. Recruit People who are good at Machine Learning to the Club Conduct Ocassional sessions that focuses on different Machine Learning Ideas.

Education

.

Kalyani Government Engineering College

Bachelor of Technology, Information Technology
August 2019 - July 2023

Current CGPA - 9.41


.

St.Luke's Day School

High School
2003 - 2017

Class 10 Overall Score - 91.8 %


.

Kanchrapara Harnett English Medium School

Senior Secondary
2017 - 2019

Class 12 Overall Score - 84 %

Skills

Programming and Other Languages
Libraries and Skills
Machine Learning 9.5/10

Computer Vision 8.5/10

Natural Language Processing 5/10

Pytorch 8/10

Sci-Kit Learn 9/10

Numpy/Pandas 9.5/10

Spring Boot 9/10

Flask 8/10

React 8/10

Notebooks, Blogs and Projects

I am a guy who loves to participate in competitions and hackathons. In my free time I develop simple website with data analysis or machine learning implementation in it. The websites are mostly developed with the help of flask, rarely I use Django as well. Here I have displayed a few solutions from some of the hackathons that I have participated in.

Notebook

This is the notebook for the submission of the TPS competition in kaggle. I got a top 15 % finish with this.

Visit The Notebook
Notebook

This is a notebook where I have used transfer learning with VGG16 on the Cat and Dog dataset

Visit The Notebook
Notebook

My notebook on the Kaggle competition to detect gravitational wave signal from detector noise.

Visit The Notebook
Blog

Here, I have discussed everything about Logistic regression. It contains everything from theory to code implementation and also got a great feedback.

Visit The Blog
Blog

This is a Technical Blog that I wrote about outliers that tries to explain what are outliers are including its mathematics and intuition and how they impact predictions

Visit The Blog
Blog

This blog is written for absolute beginners, giving them an idea about what are the important concepts that one needs to understand to learn machine learning.

Visit The Blog
Blog

This is a project where I have scrapped some economical data of India from the Internet using beautiful soup

Check The Github Repo
Project

This is the submission repository for the evaluation for the project of Deeplense of organization ML4SCI

Visti The Github Repo
Project

This is still a on going project where I try to visualize different aspects of Indian states and India as a whole

Check The Github Repo

Awards & Certifications

  • Open Source Contributor to Microsoft/LightGBM,Pytorch-Ignite/code-generator,Pycaret and more
  • Top 15% finish ( Rank 214/1495 ) in Kaggle Tabular Playground Series - March 2021 (Solo Participation)
  • Private Leader Rank of 29 and Public Leader Board Score of 306 in Analytics Vidhya Hackathon - Jobathon ( Solo Participation )
  • Top 10% finish ( Rank 146/1171 ) in Kaggle Tabular Playground Series - June 2021 (Team Participation)
  • Currently Kaggle 2x Expert ( Notebooks and Discussion )