AI/ML Researcher

SVNIT (NIT Surat) · Surat, India

2024.05 – 2024.08 · 4 mos

Research on EEG-based depression detection using deep learning architectures. I mainly work with AWS to ship useful AI products and workflow automation.

AI/ML Researcher at SVNIT (NIT Surat) focused on shipping production systems with end-to-end ownership.

I handled solution design, implementation, and delivery while collaborating with cross-functional teams to align engineering execution with business goals.

Scope: Research on EEG-based depression detection using deep learning architectures.
Ownership: architecture, implementation, and rollout
Context: high-impact business workflows and reliability requirements

Project 1

Business problem: Manual or slow workflow was limiting business throughput.
My contribution: Designed CNN-LSTM hybrid architecture for EEG-based depression detection achieving 90% accuracy — 15% improvement over state-of-the-art baselines.
Technologies: Python · TensorFlow · AWS EC2 · CNN-LSTM · Jupyter
Outcome: Delivered measurable reliability and speed improvements.

Automation Pipeline

Business problem: Manual or slow workflow was limiting business throughput.
My contribution: Engineered distributed TensorFlow training pipeline on AWS EC2, reducing training time from 12 hours to 7 hours.
Technologies: Python · TensorFlow · AWS EC2 · CNN-LSTM · Jupyter
Outcome: Delivered measurable reliability and speed improvements.

Challenge

Designed CNN-LSTM hybrid architecture for EEG-based depression detection achieving 90% accuracy — 15% improvement over state-of-the-art baselines.

Constraints

Production reliability and latency expectations
Cross-team coordination and evolving requirements

Solution

Designed pragmatic architecture tradeoffs to balance quality, speed, and maintainability.

Outcome

Shipped stable systems with measurable operational improvements.

AI/ML Researcher

SVNIT (NIT Surat) · Surat, India

2024.05 – 2024.08 · 4 mos

Research on EEG-based depression detection using deep learning architectures. I mainly work with AWS to ship useful AI products and workflow automation.

AI/ML Researcher at SVNIT (NIT Surat) focused on shipping production systems with end-to-end ownership.

I handled solution design, implementation, and delivery while collaborating with cross-functional teams to align engineering execution with business goals.

Scope: Research on EEG-based depression detection using deep learning architectures.
Ownership: architecture, implementation, and rollout
Context: high-impact business workflows and reliability requirements

Project 1

Business problem: Manual or slow workflow was limiting business throughput.
My contribution: Designed CNN-LSTM hybrid architecture for EEG-based depression detection achieving 90% accuracy — 15% improvement over state-of-the-art baselines.
Technologies: Python · TensorFlow · AWS EC2 · CNN-LSTM · Jupyter
Outcome: Delivered measurable reliability and speed improvements.

Automation Pipeline

Business problem: Manual or slow workflow was limiting business throughput.
My contribution: Engineered distributed TensorFlow training pipeline on AWS EC2, reducing training time from 12 hours to 7 hours.
Technologies: Python · TensorFlow · AWS EC2 · CNN-LSTM · Jupyter
Outcome: Delivered measurable reliability and speed improvements.

Challenge

Designed CNN-LSTM hybrid architecture for EEG-based depression detection achieving 90% accuracy — 15% improvement over state-of-the-art baselines.

Constraints

Production reliability and latency expectations
Cross-team coordination and evolving requirements

Solution

Designed pragmatic architecture tradeoffs to balance quality, speed, and maintainability.

Outcome

Shipped stable systems with measurable operational improvements.

AI/ML Researcher

Project 1

Automation Pipeline

AI

Backend

Cloud

Infrastructure

Tools

AI/ML Researcher

Project 1

Automation Pipeline

AI

Backend

Cloud

Infrastructure

Tools