• Social Robotics Programmer - Researcher @ Norwegian University of Science and Technology, Trondheim, Norway
  Jan 2025 - April 2025

  • Developed a conversational agent using Adaptive, Corrective, and Self RAG patterns, boosting accuracy by 15%
  • Integrated a Python agent with custom Kotlin Furhat skills for NorwAI’s first agent-based robot-connected backend
  • Optimized models with llama.ccp, reducing cloud dependency by 15% and achieving sub-8s responses
  • Led the full feature lifecycle from ideation to evaluation, aligning with NorwAI’s research innovation goals

• Software Engineer Co-op @ Cognite, Austin, Texas
  May 2024 — Sept 2024

  • Developed Cognite’s first industrial agent for the Atlas AI program, integrating tools for troubleshooting workflows
  • Boosted doc-parser’s keyword extraction accuracy to 90% using advanced embedding and cross-encoding techniques
  • Integrated Gemini model into Cognite’s doc-parser, enabling GCP users access and enhancing overall functionality
  • Implemented the Tail Generation Pattern to generate summaries, optimizing long-term memory recursively

• AI Engineer @ XCare, Toronto, Ontario
  Oct 2023 — Oct 2024

  • Fine-tuned Dense CNNs and Vision Transformers for X-ray diagnosis, achieving 90% accuracy
  • Developed a RAPTOR-AI pipeline, increasing retrieval accuracy to 95%, graded by medical professionals
  • Architected a RAG-AI pipeline delivering personalized rehabilitation info with references from medical sources
  • Wrote, presented, and published a paper on the tool at the Canadian Undergraduate Conference on AI

• Software Engineer Co-op @ Genellipse Inc., Toronto, Ontario
  Sept 2023 — Dec 2023

  • Optimized MongoDB architecture: enabling vector similarity search, enhancing data efficiency across 13 collections
  • Boosted data processing accuracy by 75% with Adobe and RAG, while decreasing runtime to sub-3 minutes
  • Implemented MNN and RNN Pytorch models, leading to a R2 of 0.85 and 0.95, respectively

• Full Stack Developer @ Approva Financial, Toronto, Ontario
  Jan 2023 — Dec 2023

  • Helped to secure round 2 funding within Techstars incubator through key contributions to the MERN application
  • Aided in building a machine learning-based recommendation system for matching lenders with brokers’ applicants
  • Contributed to improving lender-applicant matchmaking accuracy through predictive analytics integration

• Core Member - AI Developer @ WAT.ai, Waterloo, Ontario
  Oct 2023 — Oct 2024 Xray Tooling Project:
  • Integrated OpenAI's, HuggingFace's, and Cohere's models into the RAG pipeline, improving interpretations
  • Introduced ChromaDB into RAG pipeline, boosting rehabilitation recommendation system accuracy by 80%
  • Developed an API-driven Xray Tooling Chatbot leveraging RAG and NLP, decreasing latency to sub-2 minutes

• Languages: Java, Python, Kotlin, JavaScript, SQL, R, C#
• Frameworks: LangChain, LangGraph, Llama.ccp, HuggingFace, OpenAI, Cohere, JUnit, Express.js, Redux
• Libraries: PyTorch, pandas, NumPy, Scikit-learn, React.js, Node.js
• Tools: Docker, Firebase, Azure, AWS, Git, MySQL, MongoDB, ChromaDB

Capstone Project for MSE 541

• Tech Stack: Python
• Developed a search engine from scratch for MSE 541, leveraging BM25 for document retrieval and cosine similarity to generate query-biased summaries. Conducted retrieval analysis using hypothesis testing and t-tests to compare search engine performance and applied ranked retrieval techniques to optimize relevance and user satisfaction.

Tool Used to Help Complete MSE 343

• Tech Stack: Python, OpenAI, HuggingFace
• Developed an AI-powered audio transcription tool integrating OpenAI's GPT and Whisper models with open-source models from Hugging Face. Combined voice recognition and LLMs to parse and clean data, and implemented a human-in-the-loop system to enhance transcription accuracy and efficiency.

Capstone Project for MSCI 342

• Tech Stack: MySQL, Firebase, JavaScript, Node.js, React.js, Redux, Express.js
• Developed a full-stack web application that allows users to plan meals based on dietary preferences and allergies, generate shopping lists, and track nutritional info.

Capstone Project for MSCI 446

• Tech Stack: MongoDB, Python, Scikit-learn, PyTorch
• Applied machine learning techniques (Random Forest, XGBoost, LSTM) to predict energy prices in the US PJM Energy market, achieving significant accuracy with the Decision Tree model.

Final Project for MSCI 245

• Tech Stack: MySQL, JavaScript, React.js, Node.js, Express.js
• Built a full-stack clone of IMDB, leveraging React.js for the front end and Node.js for server-side development.

Self-Directed Learning Project

• Tech Stack: Python, Scikit-learn, MySQL
• Developed an ML model to predict player performance in the NBA based on historical data, applying algorithms such as Simple Linear Regression, K-Nearest Neighbors, and Decision Tree Regressor.

University of Waterloo
Bachelor of Applied Science (Honours Co-op)
Management Engineering, Artificial Intelligence Option
Sept 2021 — Present

• Key Courses: Machine Learning (MSCI 446), Principles of Software Engineering (MSCI 342), Databases & Software Design (MSCI 245), Algorithms & Data Structures (MSCI 240), Human-Computer Interaction (MSE 343), Search Engines (MSE 541)

• GitHub: jeevanp03
• LinkedIn: Jeevan Parmar
• Instagram: jeevan.prmr
• Email: [email protected]

Feel free to explore my projects and get in touch if you'd like to collaborate or discuss opportunities!