Ved Prakash
Pathak
Open to Work
AI|ML System Architect specializing in document intelligence, multimodal AI platforms, and production-grade LLM systems. Building AI-first products that extract value from unstructured data at scale.
Who I Am
AI|ML System Architect & Builder
Currently VP at Legend's Crafts (DocExtract.ai), leading AI product strategy for an intelligent document processing platform — extracting structured data from PDFs, invoices, and scanned images at scale.
I specialize in LLMs, RAG pipelines, Computer Vision, and industrial AI — turning unstructured data into production-grade intelligence. Open to impactful roles and collaborations.
Where I've Worked
Vice President
- Leading product strategy and AI development for DocExtract.ai, an AI-powered document intelligence platform extracting structured data from PDFs, scanned images, and invoices in Excel, CSV, and JSON formats.
- Architected an intelligent document processing pipeline integrating OCR, layout analysis, and LLM-based field extraction enabling seamless ERP and CRM integrations.
- Overseeing end-to-end platform development including invoice parsing, automated data entry elimination, custom field mapping, batch processing, and audit trails for compliance.
- Driving platform adoption and business strategy for SMEs, finance teams, and operations workflows requiring high-accuracy document automation at scale.
AI Developer
- Designed and built Intellifind.ai, a cutting-edge multimodal AI platform empowering SMBs to manage and interact with business data across spreadsheets, images, videos, and PDFs — without any technical expertise.
- Developed customizable AI workspaces for Knowledge Database, Inventory management, and Invoice processing with real-time data interaction via a conversational AI chatbot.
- Implemented RAG-based retrieval pipeline for document Q&A, enabling automated inventory updates and CSV/JSON invoice conversion with multi-format file support.
Freelance AI Developer & Machine Learning Engineer
- Built a sub-millimetre precision Computer Vision system capable of detecting hairline cracks as small as 0.05mm on 1mm rivets — enabling automated pass/fail quality control on high-throughput industrial conveyor lines.
- Engineered a real-time wire harness measurement system using ArUco marker-based spatial calibration and morphological skeletonization to measure wire lengths and detect branch junctions with high accuracy, replacing manual inspection entirely.
Independent Ventures
RactoGateway
Authored and published RactoGateway, a Python library designed to streamline AI/ML API gateway patterns, simplifying routing, authentication, and request management for AI-powered backends.
EM-telligence.ai
Offline Email Intelligence Platform leveraging locally hosted LLMs for privacy-preserving email analysis, smart reply generation, and workflow automation — without cloud dependency. Integrates Ollama-hosted models (Qwen, LLaMA variants) with GGUF quantization for efficient local inference.
Building in stealthWhat I've Built
Skills & Tools
Languages
AI Domains
AI Techniques
ML Frameworks
Cloud & DevOps
LLM Tooling
Backend & APIs
Data Science
Text Representation
Education
Bachelor of Engineering — Mechanical Engineering
Rajiv Gandhi Proudyogik Vishwavidyalaya, Bhopal
Certifications
Google Cloud & Machine Learning certifications — View Google Developer Profile →
Computer Vision
Visual data analysis and interpretation on GCP.
ML Fundamentals
Foundational machine learning principles and techniques.
ML Pipelines
Designing and deploying efficient ML workflows.
Production ML
Deploying ML models into production environments.
Recommendation Systems
Designing personalized recommendation algorithms.
TensorFlow
Building and deploying deep learning models.
MLOps
End-to-end ML workflows with Airflow, Kubeflow, Vertex AI.
+More on Google Cloud
View all badges on the Google Developer Profile.
Get In Touch
Open to opportunities
Currently VP at DocExtract.ai and actively open to new opportunities in AI/ML engineering, product leadership, and system architecture. Let's build something impactful together.