I am an Active Learner seeking to expand my knowledge base in Natural Language Programming, Machine Learning and Artificial Intelligence.

Contact Information

• Phone: 3439871466
• Email: manoj2375@gmail.com
• LinkedIn: 😄
• GitHub: 🤩
• Hugging Face: 🤗

Education

• Master’s of Computer Science
  • University Of Ottawa, Canada Grade - A+ September 2022 - August 2024
  • Courses: Machine Learning, Natural Language Processing, Virtual Assistants, Data Management, Distributed Database & Transaction Processing, Knowledge Representation, E-Commerce.
• Bachelor of Engineering (B.E)
  • Visvesvaraya Technological University, India GPA: 8.23 / 10.0 August 2016 - August 2020
  • Courses: Machine Learning, Natural Language Processing, Operating Systems, Networks.

Work Experience

Machine Learning Associate

Vector Institute, Toronto, Canada

• Duration: October 2023 - January 2024
• Curated high-quality datasets and enabled data cleaning to facilitate downstream Summarization, Q&A and Information Retrieval Tasks with 60% Domain-specific and 40% open-source data.
• Created an initial LLM pipeline to experiment with parameters and model tuning process (SFT) to enhance the performance of Mistral-7B, Zephyr-7B and LLaMA-2 with the QLoRA, LoRA and DPO techniques.
• Evaluated results on benchmark datasets such as CNN/DM, XSum, SQuAlity, SQuAD, and NEWTS also employed model distillation using GPT-4 for training and evaluation.

Machine Learning Engineer Intern

BlueGuardian, London, Canada

• Duration: February 2023 - May 2023
• Preprocessed 100k large data sets of emotions using Airflow and Pyspark and used the Pegasus text augmentation technique to do data cleaning which improved the data quality and sample size by 30%.
• Proficient in analyzing and elucidating deep learning models and NLP solutions in functional and process design, prototyping, testing, and collaboration with advanced engineering teams and executive leadership.
• Optimized XGBoost, and SVM models for high-performance emotion classification in the model architecture which could detect Sarcasm emotion with 85% accuracy.

Assistant System Engineer

Tata Consultancy Services, Bengaluru, India

• Duration: November 2020 - August 2022
• Product Design Interactions with business stakeholders to build new features using Java & Azure DB.
• Programmed the workflow for creating Rest-APIs for the web services in XML using Postman & Swagger.
• Followed Agile Methodology which improved the code quality and decreased the software bugs by 15%.

Skills and Abilities

Programming Languages

• Python, SQL (MySQL, OracleDB), NoSQL (MongoDB), C++, Java, JavaScript

Machine Learning Frameworks

• PyTorch, TensorFlow, Scikit-Learn, MLOps, LLMOps

Deep Learning

• Neural Networks, Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), Transfer Learning

Natural Language Processing (NLP)

• BERT, GPT-3.5, GPT-4, Llama2, Mistral, Zephyr, Phi2, Mixtral 8x7B
• Named Entity Recognition (NER), Sentiment Analysis, Text Classification, Summarization, Code Synthesis

Data Science Libraries

• NumPy, Pandas, Matplotlib, Seaborn, Plotly

Visualization Tools

• Power BI, Tableau

Feature Engineering

• Feature Extraction, Handling Missing Data, Outlier Detection

Machine Learning Algorithms

• Support Vector Machines (SVM), Decision Trees, Random Forests, Clustering (K-means), XG Boost, Ensemble Methods

Model Deployment

• Docker, AWS (S3, EC2, Lambda, Fargate, Bedrock, Runner, DynamoDB, RedShift), FastAPI, Flask, Kubernetes

Cloud Services

• AWS Cloud Services (S3, EC2, Lambda, SageMaker)
• Azure, Google Cloud (optional)

MLOps

• MLFlow
• Weights & Biases
• DagsHub

REST API Development

• FAST API
• Postman & Swagger for API documentation

Big Data Technologies

• Hadoop, Spark

Version Control

• Git

Projects

Summarization of Customer Feedback

• Web Scraped data from shopping sites, preprocessed and cleaned 80k samples.
• Applied K-means Clustering and fine-tuned FLAN-T5 model for abstractive summaries.
• Finetuned FLAN-T5 model to generate abstractive summaries of feedback and evaluated with ROUGE & BLEU.

Twitter Data ETL Project

• Developed an ETL pipeline using Apache Airflow, Pyspark, and AWS Cloud to extract tweets from Twitter.

StackOverflow Tag Prediction

• Web Scraped data from StackOverflow, preprocessed and cleaned 80k samples.
• Fine-tuned BERT Model, improving F1-Score efficiency by 5%.

Certifications

• Supervised Machine Learning: Regression and Classification, DeepLearning.AI, Stanford University, Coursera.
• Unsupervised Learning, Recommenders, Reinforcement Learning, DeepLearning.AI, Stanford University, Coursera.
• Generative AI with Large Language Models, DeepLearning.AI, Stanford University, Coursera.

Feel free to explore my projects, and don’t hesitate to reach out for any questions or collaboration opportunities.

Thank you for visiting!