Portfolio

Work Experience

SAAS LABS | Machine Learning Engineer
June 2022 - Present | Bangalore, India

Integral Member of Transformative JustcallIQ Team:
• As a Machine Learning Engineer at SAAS LABS, played a vital role in transforming the JustcallIQ conversational intelligence platform, introducing innovative features, and driving a significant increase in revenue.
• Developed a robust data collection pipeline from scratch, leveraging advanced technology stacks and custom Language Model Models (LLMs) to enhance machine learning operations and platform capabilities.
• Implemented AI tools like SMS Co-pilot and Context-AI, resulting in a 100% increase in customer engagement and improved workflows for customer-facing agents.
• Achieved a 75% reduction in admin review duration through the integration of conversational intelligence products.
• Enhanced communication quality by aligning client scripts with industry standards through the development of a Script Matching and Generation tool.

SAAS LABS | Machine Learning Intern
Jan 2022 - June 2022 | Noida, India

Deep Learning Intern | Guide Prof Dr.S.K.Singh | IIT BHU
Intern
• Developed a deep learning model for predicting Covid cases and segmentation of the Covid region.
• Employed Transfer Learning technique to extract features from pre-trained networks like ResNet50, InceptionResNetV2, and EfficientNet.
• Adapted different classification algorithms such as SVC, Random Forest, XGBoost, and Neural Network.
• Used data set of more than 10,000 pictures and obtained 99.9% accuracy with X-ray images and 99.4% with CT scans. The Covid region in the lungs was segmented with a 94% accuracy rate.
• The model has a greater recall and precision than the current RT-PCR test in predicting COVID cases.

Statistical Physics Data Analyst | Guide Prof Dr.Shraddha Mishra | IIT BHU
Lab Member
• Proficient in advanced statistical analysis, hypothesis testing, and data-driven decision-making, with a focus on statistical physics principles.
• Expert in applying Markov models and stochastic processes to predict and model complex, dynamic systems, essential for various applications.
• In-depth knowledge and practical experience in Bayesian modeling, providing a strong foundation for probabilistic thinking and statistical modeling.
• Skilled in developing classification models, cluster analysis, and crafting insightful visualizations, utilizing Python for efficient data processing and model development.
• Proficient in Linux-based environments, distributed computing, and simulation execution, enhancing efficiency for large-scale data analysis and insightful simulation-driven conclusions.

Datalogy
Data Science Intern
• Assisted the firm with Customer Targeting and Marketing using the insights extracted from the data.
• Performed an in-depth exploratory data analysis in Python using the NumPy, Pandas and Seaborn packages to discover the major characteristics impacting the product.
• Deployed several ML models like Linear regression, CART to predict the sustainability of the product.
• The company has experienced a 25% increase in customers using the insights.

JNCASR Bangalore
Research Intern
• Worked on a Statistical Mechanics problem using scipy, numpy, pandas, matplotlib libraries in Python under professor Subir K Das.
• Wrote a Scientific Model to efficiently predict the phase separation mechanism of a mixture when quenched to a low temperature using Mean Field Approximation of Ising Models.
• Achieved two phase separation mechanisms Spinodal Decomposition and Nucleation and employed Monte Carlo Algorithm to attain better performance.

Shreyansh Singh LLC
Marketing Intern
• Assisted with the setup and management of client presence on social sites including Facebook, and other similar platforms.
• Brainstormed ideas for new ways to engage and reach consumers.
• Increased the number of Facebook followers by 200% over 1 month period with teammates.

PROJECTS

Unpaired Image to Image Translation with Pix2Pix Generative Adversarial Networks
The Pix2PixCGAN was built using U-Net architecture, and 30x30 patch GAN architecture. This CGAN converts raw satellite images to google map style images.
Review Classification of Amazon Fine Food using Decision Tree and Random Forest
Decision Tree Classifier and random Forest Classifier were learned and the least error model was chosen by testing the model with varying hyperparameters.
Image captioning with Attention
Implemented an encoder-decoder sequence-to-sequence model with Bahdanau’s attention for translation of Spanish sentences to English sentences.
Generative Chat bot
Developed a chatbot that generates output from the input text. The graphical user interface was built using the Tkinter library in Python.
Art generation using Neural Style Transfer
The trained VGG19 model was used. The NST algorithm is used to transform content pictures into well-known art images.
Other Projects
1. Built a deep convolution neural network from scratch to classify the MNIST handwritten digits dataset.
2. Implemented a deep neural network for traffic sign classification with an accuracy of 95 percent.

PUBLICATIONS

COURSES

MLOps Specialization - Deep Learning.ai - Coursera
Business Statistics and Analysis Specialization - Rice University - Coursera
Deep Learning Specialization - Deep Learning.ai - Coursera
Machine Learning - Stanford University - Coursera
Tableau for Data Science - Udemy
Python Developer Course - Udemy

COMPETITIONS

Cascade Cup
Finished in Top 80 from 2500 participants which is jointly organized by Trell and IIT Guwahati.
ZS Campus Beats
Implemented different analytical technquies to gain inshights
Data Wiz
Finished in Top 10 from 500 participants

SKILLS

Natural Language Processing - Langchain, OpenAI, PyTorch, Transformers, GPT, VectorDB, Haystack
Python Libraries - NumPy, Pandas, Matplotlib, Seaborn, Sklearn, SciPy, Flask, JWT, FastAPI
Infrastructure and - Tools Docker, Git, Kubernetes, GCP, Sentry, MongoDB, VM, Redis, Grafana