Parth Gaikwad

Results-driven B.Tech student in Robotics & Automation Engineering (with AI & ML specialization) at Symbiosis Institute of Technology, Pune, currently in my third year. I bring hands-on experience designing, building, and deploying intelligent autonomous systems that integrate machine learning, computer vision, large language models, and real-time embedded control. My work spans the full engineering stack — from 3D mechanical design and firmware development to Python-based data pipelines, API integrations, and cloud deployments.

I have a strong foundation in Python programming, data analysis (pandas, NumPy, Scikit-learn), SQL, and data visualization (Matplotlib, Plotly, Excel). I apply an analytical and root-cause problem-solving mindset to identify process gaps, automate repetitive workflows, and deliver measurable efficiency improvements. Recognized for building production-grade systems independently, collaborating effectively in cross-functional teams, and documenting work rigorously.

Voice-Controlled 4-DOF Robotic Arm with LLM-Based AI Intent Parsing Jan 2026 – Jun 2026

Designed, built, and deployed a fully functional intelligent robotic arm capable of interpreting natural language voice commands and executing precise pick-and-place operations — combining mechanical design, embedded firmware, computer vision, and large language model integration into a single production-grade system.

• Conceptualized and 3D-modelled the complete 4-DOF arm structure in Fusion 360; designed joint geometry, link lengths, and gripper mechanism optimized for pick-and-place operations across a defined workspace envelope
• Implemented full hardware-software stack from scratch: custom 3D-printed frame, metal gear servo motors, PCA9685 PWM driver board, and external regulated PSU providing stable 5V/6V to servos under load
• Integrated Google Gemini LLM API for natural language intent parsing — the system interprets spoken commands such as 'pick up the red cube and place it on the left side' and decomposes them into structured joint-space motion commands
• Built real-time object detection and localization pipeline using OpenCV (HSV color segmentation + contour analysis) to identify target objects, compute 2D centroid positions, and map pixel coordinates to robot workspace coordinates
• Deployed dual-controller architecture: Raspberry Pi 5 handles LLM inference, vision pipeline, and high-level task orchestration; ESP32-S3 manages low-level servo PWM timing with sub-millisecond precision
• Identified a process bottleneck (redundant LLM re-initialization per command) through systematic profiling; refactored to persistent session architecture, reducing average command-to-motion latency by ~30%
• Documented full system architecture, wiring diagrams, API integration logic, test results, and calibration procedures in a structured technical report; maintained clean version-controlled codebase on GitHub

Autonomous Fire Detection & Suppression Robot Aug 2025 – Dec 2025

Built a fully autonomous ground robot capable of detecting fire hazards in real-time, computing the optimal path to the fire source, navigating around obstacles, and activating a suppression mechanism — all without human intervention. Focused on applying classical AI planning algorithms to a real embedded system.

• Developed a real-time fire detection module using OpenCV with HSV color-space segmentation; tuned detection thresholds empirically across different lighting conditions to achieve reliable identification with minimal false positives
• Implemented A* pathfinding algorithm on a 10×10 virtual grid representation of the environment; the algorithm computes shortest obstacle-free routes from current robot position to detected fire coordinates
• Built the autonomous decision loop: camera feed → fire detection → coordinate extraction → A* path computation → motor command generation → navigation → suppression trigger; entire cycle executes in under 500ms
• Applied root-cause analysis to false-positive detection events; introduced statistical thresholding (minimum contour area, aspect ratio filter) which reduced erroneous fire triggers by approximately 40%
• Integrated ultrasonic distance sensors for real-time obstacle detection and dynamic grid updates; robot re-plans path when new obstacles are detected mid-route
• Documented process flow diagrams, sensor calibration data, algorithm performance benchmarks, and test case results; produced a final report summarizing design decisions and recommendations for further improvement

Bachelor of Technology (B.Tech) - Robotics & Automation Engineering - Symbiosis Institute of Technology (SIT), Pune, Maharashtra (2023-07 - 2027-06)