Ayush Dubey

Ayush Dubey

About

Detail

Emerging AI/ML Engineer | Built end-to-end Production Systems at DRDO & Patanjali | Agentic AI · LLM · RAG · MLOps | B.E. CSE ’26
Chandigarh, India

Contact Ayush regarding: 
work
Full-time jobs

Timeline


work
Job
school
Education
folder
Project

Résumé


Jobs verified_user 0% verified
  • Deccan AI Experts
    AI Model Evaluator
    Deccan AI Experts
    May 2026 - Current (3 months)
    • Evaluate large language model (LLM) outputs across reasoning, coding, instruction-following, and factuality tasks using structured assessment frameworks. • Provide expert human feedback, preference ranking, and quality annotations to support model post-training, alignment, and evaluation workflows. • Identify edge cases, hallucinations, and failure modes through systematic analysis of model responses and benchmark-driven evaluation. • Contribute to human-in-the-loop AI development by validating model behavior, response quality, and task completion accuracy across diverse domains.ool makes LinkedIn posting effortless!
  • DRDO Ministry of Defence Govt of India
    AI/ML Intern
    DRDO Ministry of Defence Govt of India
    Jun 2025 - Dec 2025 (7 months)
    • Designed a Physics-Informed Neural Network (PINN) in PyTorch for subsurface reconstruction, formulated PDE wave-propagation constraints as custom loss terms, iterated through data ingestion, feature engineering, and hyperparameter tuning to validate the model against held-out simulation data. • Built a synthetic training data pipeline using Blender’s Python API and ANSYS simulations, automated scene parameterisation across lighting, geometry, and material variations to generate 50K+ diverse, reproducible samples at scale, eliminating the bottleneck of manual data collection entirely. • Deployed the trained model on edge hardware for real-time inference across fused multi-sensor inputs (LiDAR, ToF, RGB), handled sensor synchronisation cons
  • DRDO Ministry of Defence Govt of India
    Research And Development Intern
    DRDO Ministry of Defence Govt of India
    Jun 2025 - Dec 2025 (7 months)
    Led data-driven R&D initiatives by designing synthetic data generation pipelines, performing exploratory data analysis (EDA), feature engineering, and statistical validation for physics-based machine learning models. Developed scalable deep learning workflows, optimized model performance through hyperparameter tuning and benchmarking, and deployed low-latency inference pipelines for multi-modal sensor data in mission-critical environments.
  • Patanjali Ayurved Limited
    AI/ML Intern
    Patanjali Ayurved Limited
    Jun 2024 - Aug 2024 (3 months)
    • Built an NLP intent classification pipeline using spaCy and Hugging Face Transformers to route customer wellness queries, designed a multi-label classification layer to handle overlapping intents, chained entity extraction with a rule-based response router, improving accuracy by 18%. Delivered as a working prototype demoed to the product team. • Developed a treatment recommendation module using scikit-learn, engineered features from customer health profiles, treatment history, and preference, applied cosine similarity-based retrieval to rank Ayurvedic care suggestions, reducing recommendation retrieval time from 2.1s to 350ms.
  • Patanjali Ayurved Limited
    Artificial Intelligence and Machine Learning Intern
    Patanjali Ayurved Limited
    Jun 2024 - Aug 2024 (3 months)
    Built and optimized production-oriented machine learning pipelines for healthcare and wellness use cases, covering data acquisition, EDA, feature engineering, statistical analysis, model development, and deployment. Implemented NLP workflows for intent classification, entity extraction, and semantic retrieval, leveraging Python, Pandas, NumPy, scikit-learn, spaCy, and Hugging Face. Developed predictive models for business and operational decision-making, including time-series forecasting using ARIMA and other statistical and machine learning techniques. Improved model performance through iterative experimentation, cross-validation, and inference optimization while ensuring scalability and business impact.
  • Toloka Annotators
    AI Training Analyst
    Toloka Annotators
    Mar 2020 - Aug 2022 (2 years 6 months)
    • Contributed to AI training and evaluation workflows through Toloka, including projects delivered via the UHRS platform. • Conducted search relevance assessment, intent classification, ranking evaluation, and content categorization for machine learning systems. • Generated high-quality human feedback signals used in dataset curation and model evaluation pipelines.
Education verified_user 0% verified
  • Chandigarh university
    Bachelor of Engineering (Honors, Artificial Intelligence and Machine Learning
    Chandigarh university
    Jan 2022 - Jan 2026 (4 years 1 month)
  • C
    Intermediate
    Creative Convent College
    Jan 2019 - Dec 2021 (3 years)
  • City Montessori School
    Matriculate
    City Montessori School
    Jan 2017 - Jan 2019 (2 years 1 month)
  • Chandigarh university
    CHANDIGARH UNIVERSITY
    Chandigarh university
Projects (professional or personal) verified_user 0% verified
  • M
    Medical Image Feature Extraction using Deep Learning and SVM
    Jan 2024 - Current (2 years 7 months)
    In our project, we utilised Deep Learning, specifically Convolutional Neural Networks (CNNs), to extract intricate features from medical images, such as X-rays and MRI scans. These features were then employed in conjunction with Support Vector Machines (SVMs) for accurate diagnosis. Our approach showcased promising results, highlighting the potential of this fusion in revolutionising medical diagnostics for improved patient care and treatment planning.
  • C
    COVID-19 Detection from Chest X-Rays using PyTorch and ResNet-18
    Oct 2023 - Dec 2023 (3 months)
    Our project focused on leveraging PyTorch, a powerful deep learning framework, in conjunction with ResNet-18 architecture to tackle the critical challenge of COVID-19 detection from chest X-ray images. By fine-tuning the ResNet-18 model on a dataset comprising chest X-rays from COVID-19 positive and negative cases, we aimed to develop an accurate and efficient diagnostic tool. Through rigorous experimentation and optimization, we successfully trained the ResNet-18 model to classify chest X-ray images with high precision, enabling reliable identification of COVID-19 cases. This project underscores the potential of deep learning and its practical application in healthcare, particularly in combating the ongoing COVID-19 pandemic by facilitatin