Aayush Lal

Aayush Lal

About

Detail

Bengaluru, Karnataka, India
Contact Aayush regarding: 
groups
Networking

Timeline

work
Job
school
Education
folder
Project

Résumé

Jobs verified_user 0% verified
  • Z
    Gen AI Prompt Engineer
    Zensar Technologies (NVIDIA Project)
    Oct 2024 - Current (1 year 9 months)
    As a Gen AI Prompt Engineer, I fine-tuned and optimized NVIDIA NeMo Megatron models for large-scale language tasks, ensuring high performance and accuracy. I designed and implemented data pipelines for efficient model training and fine-tuning, which significantly improved model scalability through distributed training and hyperparameter optimization. My research and development efforts focused on advanced prompt engineering techniques to enhance the coherence and control of large language models (LLMs). I also implemented retrieval-augmented generation (RAG) strategies to improve contextual understanding, further refining the interaction between users and AI. While my primary responsibilities centered on prompt design and optimization, I al
  • C
    Data Science Intern
    CodSoft Data Science Intern
    Mar 2024 - Apr 2024 (2 months)
    Developed and deployed machine learning models for Titanic survival prediction and sales forecasting using Python, Scikit-Learn, and Pandas.Conducted exploratory data analysis (EDA) to identify key patterns and improve feature selection.Implemented classification and regression algorithms, achieving a 15% improvement in model accuracy through rigorous hyperparameter tuning and cross-validation.Created interactive dashboards and visualizations to present insights to non-technical stakeholders.Gained hands-on experience with data preprocessing, model evaluation metrics, and real-world business problem-solving.
  • 1
    Freelance Tech Lead
    1irstStep (Freelance Platform Project)
    Aug 2021 - Oct 2022 (1 year 3 months)
    Led a team delivering AI-driven solutions and full-stack projects for startups and local businesses.Oversaw technical architecture, integrated AI features, and guided junior contributors.Managed end-to-end delivery, including client communication, design alignment, and deployment.
Education verified_user 0% verified
  • RCC Institute of Information Technology
    Bachelor of Technology in Information Technology
    RCC Institute of Information Technology
    Jan 2020 - Jan 2024 (4 years 1 month)
    8.71
Projects (professional or personal) verified_user 0% verified
  • Zensar
    Autonomous Vehicle Scene Understanding
    Zensar
    Jul 2025 - Oct 2025 (4 months)
    Assisted in building structured prompt templates for driving scenarios, pedestrian activity, and road context understanding.Contributed to evaluation frameworks by aligning model-generated outputs with safety-critical events and expected driving behaviors.Helped balance precision and recall in model responses, supporting safer and more context-aware autonomous driving applications.
  • Zensar
    Skeleton Motion Description
    Zensar
    Feb 2025 - May 2025 (4 months)
    As a Skeleton Motion Description specialist, I standardized the representation of human pose sequences in prompts, enhancing model interpretation of movement patterns. I developed structured input-output mappings for skeleton data, which significantly improved model clarity in gaming and action recognition applications. My role also involved supporting evaluation processes by analyzing the alignment of model outputs with real-world motion contexts. While I have experience with cloud platforms like Microsoft Azure and AWS, I focused on motion analysis and computer vision in this position, ensuring that my skills were accurately aligned with the core responsibilities of the role. This approach allowed me to concentrate on the essential aspect
  • Zensar
    Dense Video Captioning
    Zensar
    Oct 2024 - Feb 2025 (5 months)
    Designed and refined prompt structures to guide LLMs in generating accurate and unambiguous captions from video sequences.Collaborated with the team to reduce hallucinations and improve precision by aligning outputs with expected video events.Contributed to evaluation workflows by comparing generated captions against ground-truth annotations, ensuring higher reliability.
  • H
    Harassment Detector 2.0
    Jul 2023 - Jun 2024 (1 year)
    Developed an NLP-based harassment detection system that identifies and classifies offensive text with high contextual accuracy.Applied advanced preprocessing (tokenization, embeddings, stopword removal) and fine-tuned transformer models (BERT/RoBERTa) for robust classification.Improved model precision, recall, and F1 score through hyperparameter tuning and ensemble experimentation.Built an end-to-end ML pipeline covering data ingestion, model training, evaluation, and inference.Deployed the model as a web application hosted on Netlify, enabling real-time user interaction and scalable accessibility.Released open-source implementation to promote reproducibility and future research in safe online communication.