Machine Learning Engineer

Tarush Singh

+1-513-***-**** - Portfolio - *****.*****@************.*** - linkedin.com/in/tarush-singh-246144113/ - github.com/TarushS-1996 - Medium

SUMMARY

Experienced Machine Learning Engineer with 3+ years of expertise in NLP, Computer Vision, Deep Learning, and ML/DL Optimization. Proven track record in delivering innovative solutions and driving productivity through streamlined workflows.

EDUCATION

Northeastern University Boston, USA

Masters of Science in Information Systems Sep 2022 - Apr 2024

• Relevant Courses: Neural Network Architecture, High Performance Parallel ML, Data Science engineering SRM Institute of Science and Technology Chennai, India Electronics and Communication engineering Aug 2014 - Apr 2018

• Relevant Courses: Intro to Robotics, Data structures and Algorithms, Soft Computing TECHNICAL SKILLS

Programming Languages: Python, C/C++, Java, HTML, SQL, Matlab Cloud & Tools: AWS (Lambda, EC2, S3), Azure (Functions), Git, Docker, GCP (GCF), Postman Libraries & Frameworks: Tensorflow, PyTorch, Scikit-learn, Pandas, NumPy, OpenCV, Flask, Jupyter Note- book, W&B

ML Architectures: CNN (YOLO, ResNet, Inception), Transformers (Vision Transformers, BERT), RNN

(LSTM, RCNN), ML (SVM, KNN, Decision Tree, XGBoost), AI Agents (OpenAI, Langchain, MCP)

WORK EXPERIENCE

AI Engineer

Modlee (Remote) Jul 2024 – Apr 2025

• RAG Agent Pipeline: Developed a multi-stage AI agent pipeline for retrieval-augmented generation (RAG) using Milvus DB, enriched via a tagging system.

• Boosted extraction accuracy from 65–70% to 86–95% for context-relevant queries using chunk tagging to achieve consistent, structured LLM outputs for user querries.

• API-Aware Agents: Built AI Agents designed to communicate via REST using Cube API by converting user query to REST commands

• Devised statistical and clustering-based pre-processing on returned data, reducing token usage by 75% while enabling complete summaries for 1000+ row datasets.

• Integrated distilled models for performance-efficient agent deployment, reducing token usage while preserving response quality and consistency.

NLP Development Engineer

HappSales pvt ltd, Bengaluru, India Sep 2019 - Apr 2022

• Spearheaded a NER system powered by RASA, Word2Vec and BERT, enabling voice-driven CRUD operations within CRM systems, streamlining data management and reducing processing time by 30%.

• Engineered intuitive interfaces intertwining speech-to-text, resulting in a 20% increase in user engagement and a 15% reduction in user error rates.

• Integrated the backend NLP engine with AWS Lambda for efficient and scalable hosting going from 100+ users to as required.

• Streamlined NLP integration by 40%, driving product excellence and cross-functional collaboration.

• Contributed to a 10-15% increase in user productivity by streamlining workflows and minimizing clicksDemo PROJECTS

Financial Insights RAG Feb 2024 - Mar 2024

• Led development of RAG LLM system, leveraging OpenAI’s LLM and LlamaIndex tool with streamlit for intuitive UI.

• Achieved 20% increase in data retrieval efficiency, handling diverse data sources (PDF or Snowflake DB). GitHub Multi-GPU audio sentiment analysis Jan 2024 - Apr 2024

• Engineered multi-GPU training pipeline with PyTorch’s DDP, boosting speed by 40% and emotion detection accuracy to 70%

• Deployed optimized model with TensorRT and ONNX for real-time psychological impact assessment at scale GitHub

Weight&Biases Model Analysis Jan 2024 - Feb 2024

• Explored Weight&Biases for deep learning, resulting in a 10% increase in model accuracy for smile detection in images.

• Optimized CNN model to 92% accuracy using the sweep method and provided tutorial article. Medium Model Evaluation and Parameter Mapping Feb 2023 - Apr 2023

• Conducted exploratory data analysis on attrition datasets, comparing decision tree, naive Bayes, and XGBoost models (82% accuracy).

• Improved model predictive behavior to 94% accuracy using SHAP and authored findings. Medium

Contact this candidate