Hi, I'm Deon Levon Dmello AI Engineer

AI Engineer specializing in computer vision, sensor fusion, real-time action recognition, and scalable ML systems. Experienced in deploying deep learning models with PyTorch and ROS2, integrating vision-language models and large language models, building FastAPI services, managing site reliability through SRE practices, and developing full-stack applications with AI-powered features. Passionate about creating reliable, high-performance AI solutions that bridge research and production.

Hire me
Download Resume

Experience

AI Engineer

Tardid Technologies

June 2024 - Present
  • LiDAR–Camera Calibration: Developed an automated tool for extrinsic calibration between LiDAR and camera sensors, improving spatial alignment accuracy and reducing manual calibration effort.
  • Real-time & Zero-Shot Action Recognition: Developed and deployed action recognition systems using X3D, Swin3D, and X-CLIP for zero-shot inference, achieving 30 FPS on live RTSP streams.
  • Vision-Language Models (VLMs): Integrated CLIP and X-CLIP based pipelines for cross-modal image–text retrieval, zero-shot classification, and scene understanding.
  • Large Language Models (LLMs): Deployed and optimized instruction-tuned LLMs for vision-assisted analytics and system control using llama.cpp and SGL, enabling low-latency inference on edge GPUs.
  • Face Recognition & Tracking: Implemented a perimeter surveillance system using face embeddings stored in a vector database with cosine similarity for identity matching, integrated with OC-SORT for multi-person tracking.
  • Machine Learning & Deep Learning: Fine-tuned CNN architectures for object classification, achieving a 10% improvement in F1-score on real-world datasets.
  • Backend Development: Designed and maintained FastAPI-based backend APIs for ML inference, model orchestration, and data pipelines.
  • Infrastructure & Version Control: Administered a private Git (Gitea) instance, managing CI workflows and onboarding 10+ developers.
  1. C++
  2. ROS2
  3. OpenCV
  4. GStreamer
  5. PyTorch
  6. FastAPI
  7. llama.cpp
  8. SGL
  9. CLIP
  10. X-CLIP
  11. OC-SORT
  12. Vector DBs
  13. Git/Gitea
AI Engineer Illustration

SRE/Platform Engineer Intern

Practo Technologies

March 2024 - June 2024
  • Site Reliability Engineering (SRE): Managed and resolved support tickets using JIRA, connecting to servers over SSH to troubleshoot and fix issues, ensuring system reliability and uptime.
  • Platform Team Contributions: Implemented bug patches, enhanced platform stability, and developed a thorough understanding of in-house tools for sending SMS and email notifications.
  • Microservice Management: Managed backend microservices using Docker, ensuring efficient deployment, scalability, and reliability of platform services.
  • Database Operations: Performed scheduled database backups and managed production Postgres dumps to support data recovery and environment migrations.
  • Cloud Monitoring & Automation: Developed a data pipeline to collect and analyze logs from AWS CloudWatch, enabling proactive identification of errors and deployment issues.
  • Technical Proficiency: Gained hands-on experience with PagerDuty for incident management and developed a strong understanding of Docker for containerization and deployment processes.
  1. Python
  2. Shell
  3. Ansible
  4. Docker
  5. Go Lang
  6. JIRA
  7. SSH
  8. PagerDuty
  9. AWS CloudWatch
  10. PostgreSQL
  11. Microservices
  12. Monitoring
SRE Engineer Illustration

Education

BS in Programming and Data Science

Indian Institute of Technology Madras

September 2021 - Present | Chennai, TN

Completed Diploma in Data Science & Diploma in Programming

IIT Madras

B.Tech Computer Engineering (AI & ML)

Reva University

December 2020 - July 2024 | Bengaluru, KA

Specialized in Artificial Intelligence and Machine Learning

Reva University

Projects

QuizNexus

  • Full-Stack Development: Developed a full-stack quiz platform with a Flask REST API and modular Vue.js frontend for dynamic quiz creation, administration, and responsive UI design.
  • Authentication & Caching: Implemented secure role-based authentication using JWT, Redis caching for session optimization, and MailHog/SMTP for automated email verification and notifications.
  • Analytics & Performance: Integrated Chart.js for interactive performance analytics and optimized performance through lazy-loaded components and efficient SQL querying.
  • AI Integration: Built an AI-powered student advisor using LangChain and LangGraph to analyze quiz history, detect weak areas, and generate personalized study recommendations.

  1. Vue.js
  2. Flask
  3. SQLite
  4. Bootstrap
  5. JWT
  6. Redis
  7. MailHog
  8. SMTP
  9. Chart.js
  10. LangChain
  11. LangGraph
Explore this project GitHub
QuizNexus project screenshot

ImageColorization

  • Model Development: Developed a CNN-based model using PyTorch to automatically colorize grayscale images, applying deep learning concepts such as UNet, GANs, convolutional layers, and loss optimization.
  • Packaging & CLI: Packaged the model into a Python module with clear documentation for accessibility, and currently extending it into a CLI tool for enhanced usability and portability.
  • Preprocessing & Visualization: Leveraged NumPy and OpenCV for preprocessing and visualization, ensuring smooth integration into research and prototyping workflows.

  1. Python
  2. PyTorch
  3. CNN
  4. Machine Learning
Explore this project GitHub
ImageColorization project screenshot

Sync'em

  • Full-Stack Development: Built a full-stack workforce and project management platform in an Agile Scrum environment with a cross-functional 8-member team, delivering role-based portals for Admins, HR, Project Managers, and Employees.
  • Backend APIs: Designed and implemented scalable REST APIs using FastAPI and PostgreSQL ORM models, enabling secure multi-role authentication, task workflows, and performance tracking.
  • Frontend Interfaces: Developed responsive SPA interfaces using Vue.js and TailwindCSS, providing real-time project status, employee dashboards, and management analytics.
  • Asynchronous Processing: Integrated asynchronous job pipelines with Celery and Redis for automated email notifications, scheduled reports, and background processing.
  • Generative AI Integration: Implemented Generative AI assistants (LLM + RAG) to automate HR queries, onboarding workflows, and requirement clarification, significantly reducing manual support effort.
  • Project Management: Used Jira for sprint planning, backlog grooming, story tracking, and release coordination, ensuring predictable delivery and traceability across iterations.
  • Testing & Documentation: Documented APIs using OpenAPI (Swagger) and wrote Pytest test suites for critical services, improving system reliability and maintainability.

  1. FastAPI
  2. Vue.js
  3. TailwindCSS
  4. PostgreSQL
  5. Celery
  6. Redis
  7. GenAI Agents
  8. Jira
  9. GitHub
Explore this project GitHub
Sync'em project screenshot

My Toolkit

Programming Languages

Python C++ C Java JavaScript Go Bash

Databases

MySQL MongoDB PostgreSQL SQLite

Machine Learning & AI

scikit-learn TensorFlow PyTorch

Dev-Tools

Docker Jupyter Notebook Google Colab IntelliJ IDEA Neovim Zed VS Code Git

Robotics & Computer Vision

OpenCV Gazebo ROS GStreamer FFmpeg

Operating Systems

Windows Linux macOS

Tech Stack

Vue.js FastAPI Flask Celery Redis

Generative AI (GenAI) Tools

LangChain LangGraph LlamaIndex CrewAI Ollama Hugging Face MCP

Send Message

Available for hire

As an AI engineer and backend developer, I specialize in creating machine learning models, data analysis, and data visualization. I am skilled in implementing RESTful APIs and integrating machine learning models into backend systems. Let's discuss how I can contribute to your project!

I am currently open to new opportunities and collaborations. Feel free to reach out to me for any inquiries or collaborations.

Deon Levon Dmello
Send Email
LinkedIn
GitHub