Kiran Gopi
About
Detail
AI/ML Engineer | GenAI, LLMs, LangChain, PyTorch, AWS | Ex-Founding ML Engineer @ ALL BLUE | MS CS @ Illinois Tech
Bengaluru, Karnataka, India
Résumé
Jobs
verified_user
0% verified
AI EngineerKGS Technology Group IncJul 2025 - Oct 2025 (4 months)• Designed and trained time-series deep learning models (PyTorch, TensorFlow) on large-scale sequential datasets, improving prediction accuracy by 10–15% by tuning architectures and training strategies.
• Built data pipelines using Python (Pandas, NumPy) to process raw time-series data (cleaning, normalization, windowing), improving data quality and reducing training inconsistencies.
• Developed custom evaluation metrics and validation pipelines, enabling consistent tracking of model performance and reducing regression issues across training iterations.
• Analyzed model outputs using Python, SQL, and Matplotlib, identifying error patterns and feature-level issues, leading to targeted improvements in model stability and predic- Founding ML EngineerAll BlueMar 2025 - Jun 2025 (4 months)• Built and deployed the MVP of GOJO, an AI-powered concierge platform that enables users to book flights, reserve restaurants, and book hotels through text and voice interactions.
• Developed end-to-end ML workflows using LLM APIs (GPT-based), including prompt engineering and agent-based orchestration for intent recognition, entity extraction, and task execution across booking services.
• Integrated external APIs (Yelp, flight/hotel providers, and geolocation services) to fetch real-time data and dynamically present personalized recommendations.
• Built backend orchestration logic combining LLM outputs, agent workflows, and API responses, enabling multi-step request execution and reducing manual intervention by ~40%.
• - Software Engineer InternGriffin Open SystemsJun 2022 - Jul 2022 (2 months)• Built a real-time defect detection pipeline using Flask and YOLO-based object detection models, achieving 89% detection accuracy on industrial footage and reducing manual quality inspection effort in pilot workflows.
• Prepared and annotated a dataset of 30,000+ images using Python, OpenCV, Pandas, and NumPy, improving data quality and helping reduce model false positives by 20% during experimentation.
• Developed automation modules in Java for the Griffin AI Toolkit and supported Kafka/AWS-based ML pipeline integration, contributing to faster visual inspection workflows and lower inference latency.
• Created Power BI dashboards on top of SQL-driven defect data, enabling teams to track defect trends and reducing issue resol - SResearch InternSamsung RD Institute IndiaBangaloreSRIBNov 2020 - Jun 2021 (8 months)• Designed and trained CNN-LSTM neural network models for sequence prediction, achieving 93.4% accuracy on datasets exceeding 100K samples.
• Built and curated large-scale training datasets through automated data collection, preprocessing, and normalization.
• Optimized data preprocessing pipelines using Python (Pandas, NumPy) to improve data quality and reduce training inconsistencies.
• Conducted error analysis and visualization using Matplotlib and Seaborn to guide model improvements.
• Refactored ML codebases to improve training efficiency (40% faster inference) and maintainability, enabling faster experimentation.
Education
verified_user
0% verified
- Master's degree, Computer ScienceIllinois Institute of TechnologyJan 2022 - Jan 2024 (2 years 1 month)
- Bachelor of Engineering - BE, Computer SciencePSG College of TechnologyJan 2018 - Jan 2022 (4 years 1 month)
Projects (professional or personal)
verified_user
0% verified
- IINTELLIGENT TUTORING SYSTEMAug 2023 - Dec 2023 (5 months)• Developed an AI-driven tutoring platform leveraging OpenAI GPT-4 to provide personalized learning
experiences, answering questions, and offering explanations on various subjects in real-time.
• Utilized Flask for the backend and Vue.js for the user interface, with MongoDB for storing user progress
and content.
• Deployed on Google Cloud for scalability, ensuring seamless and responsive educational support.
• Achieved a 90% user satisfaction rate based on feedback surveys, indicating high engagement and
effectiveness of the personalized learning experience. - SSmart Surveillance SystemFeb 2022 - May 2022 (4 months)• Engineered a robust Violence and Exposed Gun Detection system with a remarkable accuracy rate of
89.25%, enhancing public safety and security measures.
• Spearheaded the development of an innovative user interface, facilitating real-time surveillance data
interaction and controlled access, improving operational efficiency by 30%.
• Integrated WebSocket technology and secure user authentication, enabling seamless surveillance data
transmission and access control, leading to a 40% reduction in unauthorized access attempts.
• Applied advanced image segmentation techniques using OpenCV, including thresholding, masking, and contour-based region extraction, to isolate relevant image regions for defect detection. - EE-COMMERCE PLATFORM WITH AI-POWERED RECOMMENDATION ENGINEAug 2021 - Dec 2021 (5 months)• Led the design and development of a full-stack e-commerce platform with a machine learning-powered
recommendation engine, leveraging collaborative filtering to enhance user shopping experiences.
• Integrated Stripe for secure transactions, ensuring customer payment information is protected, resulting in a 20% increase in completed transactions.
• Utilized React, Node.js, and MongoDB for seamless frontend, backend, and database integration,
optimizing platform performance and scalability to support 1000+ concurrent users.
• Optimized database performance by implementing indexing and query optimization techniques, reducing query response times by 30%. - MMusic Recommendation System Using Emotion RecognitionFeb 2021 - Jun 2021 (5 months)➼Developed a CNN model using OpenCV and TensorFlow which involves face scanning and feature
tracking to determine the user’s mood and based on it gives a personalized playlist with a test
accuracy of 89.76%.
➼Integrated Spotify API with the VGG-16 model using FLASK to play music from user-suggested or
default playlist. - II-QuizFeb 2020 - May 2020 (4 months)➼I-Quiz is a web application where users can attend quizzes and professionals can set questions on new
topics.
➼Highlights of this application include leaving reviews for the attended quiz, rating the quizzes out of 5
stars based on the reviews, a separate tab for viewing answers of attended quiz and the details of the
user available at the profile’s page, etc. - EE-BIDJan 2019 - May 2019 (5 months)➼E-Bid is an e-commerce website for online product auctioning. It offers the user a list of the many
items up for bid in the online store. The platform offers secure bids and closing services for buyers
and sellers to encourage online bidding.
➼Sentiment Analysis approach is introduced to which emojis are presented based on the emotion and
one to one messaging system is created for the interaction between the buyer and seller.
Publications
verified_user
0% verified
- JMusic Recommendation System Using Emotion RecognitionJul