Available for Internships & Collaboration
Shivam Maurya
> Automation & Robotics Engineer · AI/ML Developer

Building intelligent systems — from autonomous ROS 2 navigation stacks to agentic AI decision engines and on-device inference pipelines.

View My Work Contact Me
40+
Embedded Projects
99.27%
ML Accuracy
6+
ROS 2 Robots
Shivam Maurya
About Me

Bridging
Robots & AI

I'm a 2nd-year B.E. Automation & Robotics Engineering student at Army Institute of Technology, Pune (2024–2028), serving as Joint Secretary of the Centre of Excellence for AI & Robotics.

My work lives at the intersection of physical systems and intelligence — I build robots that perceive, navigate, and act using ROS 2, Nav2, MoveIt 2, and on-device deep learning. I'm passionate about pushing AI out of the cloud and onto the edge.

Beyond robotics, I design agentic AI systems with LangGraph + LLMs, and independently tutor AI/ML, ROS 2, and embedded systems to 100+ peers.

Pune, India AIT · 2024–2028 Joint Sec · CoE AI & Robotics GSoC 2026 Applicant
Timeline
2024 · PRESENT
B.E. Automation & Robotics · AIT Pune
Relevant coursework: Robotics Systems, Control Theory, Computer Vision, ML, Embedded Systems, IoT & Sensor Networks.
JUN 2025 · PRESENT
Joint Secretary · CoE AI & Robotics
Co-lead 100+ member research cell. Organise workshops, hackathons, and project sprints across physical AI, CV, and embedded robotics.
2024 / 2025
ABU Robocon · AIT Team
Led CNN-based real-time vision system for competition robot. Represented AIT Centre of Excellence for AI & Robotics.
2026
GSoC 2026 · OSRF Applicant
Targeting Physical AI Inference & Trajectory Upscaling project with Open Source Robotics Foundation.
Technical Skills

My Stack

🤖
Robotics & Autonomous Systems
ROS 2 Full Stack
ROS 2 Humble Nav2 MoveIt 2 SLAM Toolbox Gazebo RViz2 URDF/XACRO tf2 ros2_control Nav2 BT OMPL DWB Planner micro-ROS
💻
Languages
Core Languages
C++ (primary)
90
Python
92
Bash
70
SQL
65
🧠
AI & Deep Learning
ML / DL Stack
TensorFlow Keras PyTorch Scikit-Learn XGBoost ONNX Runtime CNN Xception TL YOLO Pandas NumPy
👁️
Robot Perception
Sensors & Vision
OpenCV LiDAR Depth Camera Point Cloud (PCL) Sensor Fusion Camera Calibration Real-Time Inference
⚡
Agentic AI & LLMs
LLM Orchestration
LangGraph LangChain Gemini 2.5 Pro ReAct Agents RAG Pipelines Pinecone FastAPI Streamlit
📡
Embedded & IoT
Hardware Stack
Arduino ESP32 ESP8266 SIM800L GSM UART/I2C/SPI Motor Drivers MQTT Firebase
🛠️
Tools & DevOps
Dev Environment
Git / GitHub Docker CMake / colcon Linux (Ubuntu) VS Code Jupyter Google Colab
Featured Work

Projects

From autonomous robots navigating GPS-denied spaces to agentic AI systems making clinical decisions.

🤖
Robotics · Physical AI
Physical AI Inference Pipeline
↗
Modular AI perception pipeline integrated into a ROS 2 robot architecture. YOLO + CNN classifiers deployed as dedicated perception nodes, exporting TensorFlow models to ONNX for optimized on-device inference — hot-swappable backends, results consumed by Nav2 costmaps and MoveIt 2 grasp planners. Validated live on ABU Robocon competition robot.
30+ FPS
Real-time · No GPU
ONNX
Runtime backend
Closed
Perception→Action loop
ROS 2YOLOONNX RuntimeTensorFlowC++PythonNav2MoveIt 2
🚗
Navigation
Autonomous Ground Vehicle · Nav2 Full Stack
Complete Nav2 navigation stack on a differential-drive robot. SLAM Toolbox for real-time 2D occupancy mapping from LiDAR. DWB local planner + Navfn global planner with recovery behavior trees.
GPS-Free
Indoor localization & loop closure
ROS 2 HumbleNav2SLAM ToolboxLiDARDWBC++
🦾
Manipulation · Physical AI
6-Axis Robotic Arm — AI-Guided Pick & Place
Full physical AI loop: camera input → on-device CNN pose estimation → grasp target generation → MoveIt 2 Cartesian planner → physical execution via ros2_control. OMPL planners (RRT*, RRTConnect, PRM) benchmarked for joint-space and task-space planning.
6-DOF
Custom XACRO arm
RRT*
OMPL motion planning
ROS 2MoveIt 2OMPLURDF/XACROTensorFlowOpenCVros2_control
💳
Agentic AI
CREDI-MITRA — Agentic Credit Appraisal
↗
5-phase autonomous credit appraisal using LangGraph ReAct agent, Gemini 2.5 Pro, XGBoost classifier, and Pinecone RAG over financial knowledge base. Hackathon 2026 winner.
97%
XGBoost ML accuracy
LangGraphGemini 2.5XGBoostPineconeFastAPI
🔬
Healthcare AI
Antimicrobial Resistance Prediction
Random Forest classifier for IMIPENEM resistance prediction on 274 bacterial isolates. WHO/CLSI-aligned clinical decision framework. CODECURE Hackathon Track B top submission.
99.27%
F1 = 0.7982 · Stratified K-Fold CV
Scikit-LearnXGBoostRandom ForestPython
🏥
Healthcare AI
MedSynapse — Multi-Disease AI Platform
↗
Multi-disease prediction platform with 4 integrated DL models. Diabetes ~98% accuracy. Brain MRI 4-class tumor classification via Xception transfer learning. Selected for IIIT Pune Thinking Machine Competition 2025.
~98%
Diabetes model accuracy · Live on Streamlit
TensorFlowXceptionKerasStreamlit
🎙️
Voice AI
MedSarthi — Multilingual Voice Health AI
Multilingual voice health assistant with 4 diagnostic models, STT/TTS pipeline via Sarvam API, and Gemini API for clinical reasoning. On-device ML inference for rural healthcare accessibility.
Sarvam STT/TTSGemini APIOn-device MLPython
📡
Embedded IoT
NanoTracker-24/7 — Ultra-Low-Power Tracker
GPS-free asset tracker using ESP8266 LBS triangulation + SIM800L GSM. 99% deep-sleep efficiency. SMS-based alerting with cloud dashboard integration.
99%
Deep-sleep efficiency
ESP8266SIM800LLBSSMSC++
🏭
Industrial AI
Predictive Maintenance System
↗
Random Forest classifier on AI4I 2020 dataset. Dual-mode Streamlit dashboard: Manual Diagnostics + Fleet Monitoring for 9+ machines. Directly applicable to robot joint health monitoring.
>98%
Classification accuracy · Live on Streamlit
Random ForestScikit-LearnStreamlitPython
Recognition

Achievements

🥇
CODECURE Hackathon — Track B
AMR prediction system achieving 99.27% accuracy with WHO/CLSI-aligned clinical decision framework. Top submission across all teams.
🏆
IIIT Pune "Thinking Machine" 2025
MedSynapse multi-disease AI platform selected for the Diagnostic Tools track. Live-deployed with 4 integrated ML/DL models.
🤖
ABU Robocon 2024/25
Deployed CNN-based real-time vision classifier on competition robot. Represented AIT Centre of Excellence for AI & Robotics.
🏅
Hackathon 2026
CREDI-MITRA agentic AI credit appraisal system — 97% XGBoost accuracy, deployed live on Streamlit Cloud.
🎓
Joint Secretary · CoE AI & Robotics
Leading AIT's AI & Robotics research cell with 100+ members. Organising workshops, hackathons, and technical sprints.
📚
AI/ML & Robotics Tutor
Independent teaching — Python, ML, ROS 2, ESP32, and IoT. Building the next generation of robotics engineers at AIT.
Get In Touch
Let's Build Something
Intelligent.

Open to internships, research collaborations, open-source contributions, and robotics/AI projects. I respond fast.

GitHub ↗ LinkedIn ↗ Download Resume