Guddu Kamti - Computer Vision Developer

Overview

Interactive computer vision prototyper (junior) focused on real-time camera demos and gesture interfaces. The strongest proven skill is building a realtime hand-tracking air-paint demo using MediaPipe and OpenCV as implemented in tut_air_canvas.py. There is no evidence of model training, experiment tracking, production deployment, or automated tests in the public code.

Phone

Timeline

Computer Vision Engineer · Middle

Artificial Intelligence Technologies Ltd. · Full-Time

Feb 2024 to Present 2 Years 9 Months Noida In office

Built a real-time hand gesture recognition solution using webcam input, including landmark-based feature preprocessing and gesture classification. Tuned inference to reduce latency and improved stability to lower false detections. Collaborated with a cross-functional team to design, test, and deliver production-ready computer vision applications.

OpenCV

MediaPipe

Computer Vision intern · Junior

Aspagteq Technology Pvt. Ltd. · Full-Time

Jul 2023 to Dec 2023 5 Months Noida In office

Developed a facial emotion recognition pipeline for live video, including face detection and expression classification. Trained models on a large labeled dataset and integrated HCI behavior so the application could react to user state in real time. Supported testing and integration of computer vision models into the main software system.

OpenCV

Machine Learning

Computer Vision Internship · Junior

AIM Point Infosystem Pvt. Ltd. · Internship

Bhopal In office

Completed an internship focused on computer vision work and Python-based project development, contributing to practical tasks under the internship program.

Projects

Real-Time Object Detection & Tracking System

Implemented an end-to-end video pipeline for multi-object detection and tracking using a YOLO-based model. Added tracking so objects maintain consistent IDs across frames while processing live video. Exposed the model via a lightweight Flask API for serving real-time predictions.

Hand Gesture Recognition System

Built a real-time hand gesture recognition application using webcam input for touchless control. Used landmark-based feature extraction to classify multiple gestures with high observed accuracy. Focused on making the gesture recognition responsive for interactive use cases.

CCTV Surveillance System

Created a real-time surveillance application that detects and tracks people and vehicles on live camera feeds. Implemented event alerts for restricted-zone entries and stored timestamped snapshots for later review. Delivered a simple web dashboard that serves the live feed and alert notifications.

Technical skills

Python

Flask

AI/ML

OpenCV

MediaPipe

Computer Vision

Machine Learning

NumPy

YOLO

PyTorch

Vision

DevOps

GitHub

Git

Rest API

Junior AI/ML Engineer Confidence: Low Generalist

Interactive computer vision prototyper (junior) focused on real-time camera demos and gesture interfaces. The strongest proven skill is building a realtime hand-tracking air-paint demo using MediaPipe and OpenCV as implemented in tut_air_canvas.py. There is no evidence of model training, experiment tracking, production deployment, or automated tests in the public code.

Model Architecture & Training

How well models are designed and trained

Not evidenced in public code

Data Pipeline & Feature Engineering

2/10

How data is prepared for models

Basic real-time image preprocessing and landmark extraction for hand tracking and gesture-based input.

Evidence

gudduk01/shadow-remove/tut_air_canvas.py: framergb = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB) gudduk01/shadow-remove/tut_air_canvas.py: building landmarks list from result.multi_hand_landmarks and mapping normalized coords to pixel coords

Experimentation & Evaluation

How results are measured and tested

Not evidenced in public code

MLOps & Deployment

How models are shipped to production

Not evidenced in public code

Computational Efficiency

2/10

How efficiently computing resources are used

Simple runtime considerations for bounded memory and real-time drawing using deque with fixed maxlen and incremental drawing to avoid large allocations.

Evidence

gudduk01/shadow-remove/tut_air_canvas.py: use of collections.deque(maxlen=1024) for point buffers gudduk01/shadow-remove/tut_air_canvas.py: incremental cv2.line drawing to frame and paintWindow to maintain real-time responsiveness

Research Depth & Innovation

Depth of research and new ideas

Not evidenced in public code

Expertise

Document Intelligence & OCR· Junior

Technologies

Vision

Python· since 2026 · Junior

OpenCV· since 2023

YOLO

Computer Vision· since 2023

NumPy

Git

PyTorch· since 2026

MediaPipe· since 2024

GitHub

Machine Learning· since 2023

OCR· mentioned only

Recommendations

Develop rapid prototypes of camera-based UX and gesture-controlled demos using OpenCV and MediaPipe.
Build preprocessing pipelines for shadow reduction and OCR-ready image cleanup, leveraging current frame processing code as a base.
Harden code for production by adding error handling, configuration, and simple unit or integration tests around capture and processing loops.

Repositories

The developer's experience in this domain has been verified based on AI analysis of the following repositories:

shadow-remove

Job preferences

Work permits: India
Employment type: Full-Time
Locations: Noida
Company size: 1-10