Ashok Kumar

AI/ML engineer with startup experience driving agentic AI systems, LLM platforms, computer vision models, and recommendation engines from design through production deployment.

AI Engineer - Vienstereoptic Pvt. Ltd.

(2025-10 - 2026-07)

• Led AI implementation strategy and system design across all new and existing company projects, setting technical direction for the AI function.
• Designed and shipped Agentic AI and RAG / CRAG pipelines using LangChain and LangGraph, including hyperparameter tuning to improve model accuracy and response quality.
• Fine-tuned LLMs on GPU/CUDA infrastructure and owned end-to-end integration and deployment of models into live production systems.
• Containerized and deployed AI services with Docker; set up local LLM environments (Hugging Face, Ollama) sized to project and hardware requirements.
• Built vector database and embedding-model pipelines to power retrieval-augmented generation across multiple products; used LangSmith for LLM tracing and evaluation.
• Used Playwright for web automation, including a GenAI content pipeline and image-generation tooling; generated synthetic demo datasets with Faker for realistic project demos.
• Built an internal AI-as-a-Service (AIaaS) platform issuing API keys so company developers could integrate AI models into their own projects with minimal setup; worked hands-on with multi-tenancy and microservices patterns.
• Led the AI Digital Marketing project, an NRI face-verification project, and an AI-based grievance-handling system.
• Mentored summer interns end-to-end on AI project planning, debugging, AI awareness sessions, and system design guidance.
• Administered Linux and company server VMs, including local LLM installs configured to each VM's system requirements.
• Built and deployed a dynamic LLM-powered chatbot using RAG for the company website, enabling context-aware responses from internal data sources.
• Led AI integration for a School Management System, designing and integrating multiple ML models to support academic analytics and administrative automation.
• Developed a fabric defect detection system using YOLO-based object detection, covering dataset preparation, model training, and inference testing.
• Built an HRMS ATS and recommendation engine to assist recruiter shortlisting using similarity-based ranking techniques.
• Implemented AI features for an e-learning platform, including live speech-to-text transcription, automated notes, MCQ generation, and a doubt-resolution chatbot.

B.E. - Computer Science and Engineering - St. Mother Theresa Engineering College (2021 - 2025)