Experience

• 

  traversaal.ai

  • AI Researcher Co-op

    • Mar 2024 - Present

    - Building End to End enterprise level LLM powered Co-pilots and RAG systems.- Deploying Opensource LLMs such as Mistraal on Multi-GPU systems - Utilizing Huggingface frameworks such as Deepspeed, DPO, RLHF and Bits&Bytes for finetuning LLMs

• 

  Packt

  • Technical Reviewer

    • Apr 2024 - Present

• 

  Think In Bytes

  • Ahmedabad, Gujarat, India

  • Co-Founder

    • Aug 2020 - Present
    • Ahmedabad, Gujarat, India

• 

  Santa Clara University

  • Santa Clara, California, United States

  • Research Assistant - TailorED Project

    • May 2023 - Dec 2023
    • Santa Clara, California, United States

    Role and Project:Position: Research Assistant at Santa Clara University.Project Goal: Automate fetching insights from syllabus PDF to help faculty in optimizing syllabus planning to maximize student performance through better syllabus structuring.NLP Techniques:Implemented NLP techniques.Tasks: Entity recognition, keyword extraction, text embedding.LLMs Integration:- Finetuned Llama-2 and Falcon 7B LLMs using PEFT and QLora on custom datasetPurpose: Enhance syllabus insights and correlate with student performance.

• 

  Teksun Inc

  • Ahmadabad, Gujarat, India

  • Machine Learning Engineer L2

    • Oct 2020 - Dec 2022
    • Ahmadabad, Gujarat, India

    1. Role and Responsibilities: - Position: Machine Learning Engineer at Teksun Microsys Pvt Ltd. - Focus: Developing and implementing end-to-end AI and ML pipelines.2. Lane Detection Algorithm: - Developed a State of the Art lane detection algorithm for an autonomous assisted driving system. - Techniques Used: Gradient filter, perspective transform, and custom image segmentation in PyTorch. - Result: Achieved a 96% pixel accuracy for assisted driving mode on Indian roads. - Client: Japanese Car manufacturing company with 80,000+ customers in India.3. Video Classification Model: - Designed a lightweight video classification model. - Architecture: CNN-LSTM. - Application: Accurate detection and counting of repetitions for 40 different gym exercises. - Deployment: Implemented and deployed on Nvidia Jetson nano.4. Object Detection for Hazard Detection: - Implemented YOLOv5 Object Detection for hazard detection in commercial buildings. - Contribution: Played a role in building incremental learning pipelines.5. ML Orchestration and Monitoring: - Implemented ML orchestration and monitoring using Airflow DAGs. - Application: Qualcomm 625 based commercial product. - Clientele: Served numerous B2B clients in India and the US.

• 

  HypeTeq Software Solutions Pvt. Ltd

  • Junior Software Engineer - ML/AI

    • Jul 2020 - Oct 2020

    Role and Responsibilities:Position: Junior Software Engineer (AI/ML) at Hypeteq Software Solutions Pvt. Ltd.Contribution: Actively involved in the development of energy prediction systems.Wind Energy Prediction Project:Project Focus: Wind energy prediction.- Implemented custom Time-series based stacked machine learning boosting model.Performance: Achieved a remarkable Mean Absolute Percentage Error (MAPE) of 1.40%.Benefit: Improved resource utilization in the energy production process for seimens based wind-farms across Western India.

  • Software Engineer trainee

    • Jan 2020 - Jun 2020

• 

  Udemy

  • Online Instructor

    • Jul 2017 - Nov 2019

    I am an online ethical hacking trainer at Udemy learning platform and share my hacking skills that are highly in demand. I am on my way to make technology a better and a safer thing to dwell with.

• 

  Self-employed

  • Book Author - "Hack yourself before someone else does"

    • Feb 2017 - Feb 2017

    Authored a book on ethical hacking and cyber secuirty that aims to make technology a better and a safer thing to dwell with by spreading awareness about cyber crimes and threats to information assets.