Pranav Dave

About Pranav Dave

I am an AI Software Architect and Senior Software Engineer with over a decade of experience building and scaling intelligent systems. My expertise lies at the intersection of Machine Learning, MLOps, and high-performance software engineering, with a proven track record of reducing processing times from days to hours through GPU acceleration and pipeline optimization.Currently, I serve as an AI Software Architect at Crest, where I support the Toyota Research Institute (TRI) in developing advanced MLOps pipelines and scenario-prediction models for autonomous driving. Throughout my career, I have led cross-functional teams to deliver end-to-end AI products, including Agentic GenAI workflows, automated PLC circuit analysis, and scalable HR-tech voice bots.Core Competencies:GenAI & LLMs: Agentic workflows (MCP, LangChain), RAG architecture, and custom security copilots.Computer Vision & Robotics: YOLOv8, Vision Transformers (ViT), and automated labeling via SageMaker Ground Truth.MLOps & Infrastructure: AWS/Azure deployment, Kubernetes, Terraform, and high-performance multiprocessing.Engineering Leadership: Managing teams of engineers and interns to take products from POC to production-ready SaaS.

Experience

1. Senior Software Engineer

   Crest Data

   Jul 2024 — Present · Ahmedabad, IN

   Contract work at Toyota Research Institute (TRI)- MLOps- Optimized robotic data (generated using Carla) post-processing by developing a multi-participant pipeline and accelerating FFMPEG with GPU, reducing video processing time from 7–10 days to ~15–20 hours. Leveraged multiprocessing in Python & shell scripts, cutting execution from 2–3 hours to ~15–20 minutes for faster data labeling- Designed custom data labeling jobs in SageMaker Ground Truth and explored ML model training & deployment, providing insights for DevOps integration- Developed and optimized a Dreamer-based reinforcement learning pipeline for AWS SageMaker, including end-to-end debugging, experiment automation, and performance tuning for large-scale training workloads. Additionally, built a CI pipeline to automatically validate training scripts on new PRs, ensuring reliability and preventing regressions before merging to main- Designed and evaluated a scenario-prediction model using the TRI-generated Driver Behavior Models (DBM) dataset, integrating a visual encoder–decoder architecture with a Vision Transformer (ViT) to enhance scene understanding and trajectory prediction accuracy.Sr. Software Engineer (Team Lead)- Developed an end-to-end automated system that converts PLC circuit PDFs into structured XML rule files- Led a team of 5 and built a Market Research web app for analyzing news and financial documents using advanced RAG architecture- Developed an Agentic workflow for SOC analysis, utilizing GenAI and Agentic architecture to analyze incidents, alerts, and phishing data with a chain of thought for large log analysis- Replaced existing tools with the MCP server implementation so that all the agents are framework agnostic and can communicate seamlessly- Created a custom Agentic security copilot plugin combining API, KQL, and GPT skills for SOC data analysis across Netskope, Microsoft Entra, and Purview.

Education

• Birla Institute of Technology and Science, Pilani

  Master of Technology - MTech, Data Science and Engineering

• Gujarat University

  Bachelor of Engineering - BE, Mechanical Engineering

  2005 — 2009

Skills

• Fortran
• Simulations
• Fluid Dynamics
• Mechanical Engineering
• Numerical Analysis
• Hypermesh
• Computational Fluid Dynamics (Cfd)
• Cfd
• Pro Engineer
• Autocad
• Matlab
• Catia
• Heat Transfer
• Fluid Mechanics
• C
• Openfoam
• Cfx
• Engineering