DEEPAKROSHAN
· IST
--:--:--
---
AI Engineer
Building intelligent systems
that solve real problems
deepakroshan380@gmail.com github.com/deepakroshan11
Scroll
Deepak Roshan A (PORTFOLIO)
Deepak Roshan A
AIEngineer
Deepak Roshan A
Coimbatore · India
AI.ENG · DEEPAK ROSHAN A · 2026
Tilt me ↕
01    About
Building intelligent
systems that matter

AI Engineer specialising in Python, Machine Learning, and AWS cloud infrastructure, focused on designing and deploying end-to-end intelligent systems. From training deep learning models to delivering production-ready APIs, I build systems that don't just work in theory — but scale, perform, and create real-world impact.

With a strong foundation in AI and cloud engineering, I build scalable, reliable, and production-grade systems, continuously leveraging modern tools to create future-ready solutions. Every architecture decision is made with performance, cost, and longevity in mind.

I work extensively on building, optimising, and deploying machine learning models as production-grade solutions, focusing on turning complex ideas into practical, user-centric systems — from CNNs and computer vision pipelines to self-healing cloud infrastructure and real-time AI platforms.

Building intelligent systems that move beyond code — into real-world impact.

deepak@ai-engineer ~ ● LIVE
02    Experience
Where I have
contributed
Apr 2025 — Jun 2025
Appin Technology
Cloud Operations Intern
  • Deployed applications on AWS using EC2, S3, and Lambda with Linux environments.
  • Automated deployment and system tasks using shell scripts for operational efficiency.
  • Monitored live systems and logs using cron and htop, maintaining uptime and performance.
  • Optimised cloud resource allocation for improved cost-performance across environments.
2022 — 2024
Freelance
Frontend Developer · Part-time
  • Built scalable and responsive web interfaces across multiple client projects.
  • Designed user-friendly, modern UI/UX with a focus on visual clarity and usability.
  • Developed UI components for real-world product applications from scratch.
  • Improved performance and load efficiency through code optimisation and asset management.
03    Education
Where I have
learned & grown
2022 — 2026
Sri Krishna College
of Technology
B.E. in Artificial Intelligence and Machine Learning
Current · Final Year
2020 — 2022
John Dewey Matric.
Hr. Sec. School
Higher Secondary Education · Science Stream
Completed · 2022
04    Projects
Things I have
built & shipped
01
Findora
AI-powered lost & found — reconnecting people with what matters
React · FastAPI · MobileNetV3 · MiniLM · Supabase
  • Upload a photo of a lost item — AI instantly scans and matches it against found reports using visual + text similarity
  • Smart scoring ranks results by confidence — highest matches trigger automatic email/SMS alerts
  • Location-aware search narrows matches to your area so the right people are notified fast
  • Real-time dashboard shows all reports, match scores, and resolution status at a glance
AI Match EngineImage + Text + Location
02
ISL Translation + Emotion Aware
Breaking barriers — Indian Sign Language understood in real time
TensorFlow · MediaPipe · Flask · Three.js
  • Point your camera — the system reads hand gestures live at ~95% accuracy using a trained CNN model
  • Detects your emotion through facial landmarks alongside translation, making conversation more natural
  • Type any text and a 3D animated avatar signs it back to you in ISL — bridging both directions
  • Runs entirely in the browser via Hugging Face Spaces, no installation needed
Real-time3D AvatarCNN Model
03
CloudGuard AI
Your cloud, always healthy — autonomous monitoring & self-healing
AWS · Python · Isolation Forest · Streamlit · Lambda
  • Continuously watches EC2 metrics via CloudWatch — CPU, memory, network — without you lifting a finger
  • Isolation Forest ML model detects unusual patterns before they become outages
  • Automatically triggers Lambda recovery scripts when anomalies are confirmed — true self-healing infrastructure
  • Sends instant SNS alerts with root-cause context so you always know what happened and why
Auto-HealAnomaly DetectionAWS Lambda
04
VoxAI Neural Voice Clone Studio
Your voice, anywhere — clone any accent in seconds
Python · FastAPI · ChatterboxTTS · Hugging Face
  • Upload just 3–10 seconds of any voice — the system clones it without any training or fine-tuning
  • Preserves accent, tone, and speaking rhythm so the output sounds indistinguishably natural
  • Built-in 7-stage audio enhancement pipeline cleans and polishes the synthesized voice output
  • Accessible via Hugging Face Spaces — try it instantly, no API key or setup required
Zero-shotAccent PreserveAudio Pipeline
05    Certifications
Credentials &
achievements
AWS
AWS Educate — Cloud Computing Foundations
Amazon Web Services
2025
Anthropic
Claude Code in Action
Anthropic Academy
2026
OpenCV
Introduction to Computer Vision with OpenCV
Coursera / OpenCV.org
2024
06    Skills & Stack
My tools to
build with precision

From model training to cloud deployment — every layer of the stack, owned end-to-end.

Click & drag to explore
Deepak Roshan A  ·  AI Engineer  ·  2026
Thank you for visiting my portfolio.
Designed & Developed by Deepak Roshan
07    Contact

Ready to build something that matters?
Let us connect and make it real.

Let's Connect