Open to opportunities

Mitul User

@mituluser

Message

Research-focused AI engineer specializing in LLM systems and retrieval-grounded architectures.

India

Message

What I'm looking for

I seek roles that let me build and deploy research-driven LLM and retrieval systems, work cross-functionally, and measure real-world robustness and fairness while growing in applied research and production ML engineering.

I am a research-focused AI engineer studying agentic LLM systems and retrieval-grounded generative architectures, with hands-on experience building structured multi-stage pipelines to evaluate reasoning, generalization, and failure modes under real-world constraints.

I designed and curated the PashuVision dataset of 216,243 images across 45 cattle breeds and benchmarked YOLOv8/11/26 models, achieving 97.4% Top-1 and 99.9% Top-5 accuracy while rigorously measuring calibration and fairness impacts. I build end-to-end systems—OCR, RAG with vector embeddings, prompt chains, and mobile-friendly inference—for products like Qurio and engineered real-time multiplayer AI features for Prompt Battle.

I bring production-oriented engineering skills (Python, PyTorch, LangChain, Flutter, React, ROS) combined with strong experimental rigor and a focus on robust deployment; I enjoy turning research insights into reliable, user-facing systems.

Experience

Work history, roles, and key accomplishments

Current

AI/ML Engineer (Student Project)

Current

PashuVision

Jan 2026 - Present (5 months)

Led dataset curation and benchmarking for a 216,243-image, 45-breed cattle classification dataset, achieving 97.4% Top-1 and 99.9% Top-5 accuracy using YOLO-based classification and class-weighted sampling.

Data Curation PyTorch YOLOv8 11 26 Transfer Learning McNemar)Python

Current

Research Intern

Current

ATOM Robotics Lab

Nov 2025 - Present (7 months)

Contributed to a ROS/Gazebo driving simulator project, designing CAD peripherals and integrating micro-ROS on ESP32 to enable real-time communication between simulated and physical components.

ROS Gazebo ESP32 CAD Real Time Systems Linux Git

Current

Research Engineer

Current

Independent / Academic Projects

Oct 2025 - Present (8 months)

Developed large-scale ML pipelines and retrieval-augmented generative systems for applied research, producing a 216k-image cattle-breed benchmark and RAG-powered mobile note-understanding app with LLM-driven summarization features.

PyTorch YOLOv8 11 26 RAG Gemini API Python Mobile UX

Full Stack Developer

Prompt Battle (Project)

Oct 2025 - Nov 2025 (1 month)

Built a real-time multiplayer AI game with synchronized WebSocket state, integrated Gemini API for prompt evaluation, and an SDXL image-generation pipeline for competitive gameplay.

React TypeScript Socket.IO Gemini API WebSockets Real Time Systems Stability AI

Education

Degrees, certifications, and relevant coursework

Maharaja Agrasen Institute of Technology

Bachelor of Technology, Information Technology

2024 -

Grade: 9.23 / 10

Activities and societies: Projects: PashuVision (cattle breed classification), Qurio (AI-powered note understanding), Driving Simulator (ROS/Gazebo), Prompt Battle (real-time AI game); technical leadership in ML, RAG, and embedded-ROS integration.

Pursuing a B.Tech in Information Technology with a strong academic record (CGPA 9.23/10) and focus on building agentic LLM systems, retrieval-grounded architectures, and applied ML pipelines.