Journey

A timeline of my learning, projects, and milestones.

Here is a look at my technical journey, from writing my first lines of code to developing full-stack AI applications and shipping production-ready machine learning models.

Now

  • Open Source & Community - Maintaining a consistent open-source activity streak, shipping production-ready AI applications, and growing a community of 1,000+ GitHub followers.

2026

  • June 2026 - Present - Research Intern at DRDO NPOL (Kochi), scaling underwater acoustics deep learning frameworks across massive real-world acoustic datasets.
  • May 2026 - Graduated with a B.Tech in Computer Science & Engineering (AI Engineering) from Muthoot Institute of Technology & Science (CGPA: 7.56/10).
  • Feb 2026 - Shipped SERVD, a full-stack AI kitchen assistant integrating Gemini Vision AI, Next.js, Strapi, PostgreSQL/SQLite, Clerk Auth, Arcjet, and Razorpay for automated ingredient recognition and recipe generation.

Highlights

  • Scaled the DWSTr deep learning framework to process the massive 47-hour DeepShip real-world acoustic dataset, targeting the classification of 265 distinct vessels across 4 commercial categories.
  • Engineered a memory-efficient data pipeline to systematically chunk high-resolution audio files into compressed .npz tensors, mitigating cloud Out-of-Memory (OOM) bottlenecks during scale-up.
  • Architected two separate pantry ingestion workflows (image-based scanning and manual entry) in SERVD with expiry tracking and real-time database sync.
  • Integrated end-to-end payment processing via Razorpay with role-based premium access controls and secure authentication via Clerk and Arcjet.

2025

  • Sep - Dec 2025 - Research Intern at DRDO NPOL (Kochi), architecting scalable deep learning frameworks for automated passive sonar audio classification.
  • Sep 2025 - Built SELECTRA, an AI-powered resume screening platform utilizing Claude Opus 4.1, React Router v7, TypeScript, Tailwind CSS, and PDF.js to generate ATS compatibility scores across 5 rigorous evaluation criteria.

Highlights

  • Developed the foundational hybrid DWSTr model (Depthwise Separable CNN + Transformer) for passive sonar classification, achieving a benchmark 97.17% accuracy across 12 vessel categories.
  • Implemented square root class-weight smoothing to resolve severe dataset imbalances in naval acoustic data.
  • Built the initial end-to-end audio pipeline processing 10,000+ sonar samples into Mel-spectrograms, slashing preprocessing latency by 35%.
  • Presented the DWSTr model to senior NPOL leadership, gathering actionable feedback for iterative model enhancements.
  • Integrated Puter.js (cloud storage, KV database, AI API) and Zustand state management into SELECTRA for persistent resume tracking, which improved user retention by 60%.
  • Solved 300+ problems on LeetCode, solidifying core Data Structures and Algorithms patterns alongside academic coursework.

Foundations (2023 - 2024)

  • Began programming with C, C++, Java, and Python, building a foundation in algorithmic thinking, data structures, and object-oriented design.
  • Deepened core Computer Science fundamentals — Data Structures, DBMS, Operating Systems, Computer Networks — through university coursework and independent projects.
  • Transitioned into modern web development, picking up React, Next.js, TypeScript, Tailwind CSS, REST APIs, and full-stack architecture.
  • Started exploring Machine Learning and Data Science with Python, NumPy, Pandas, Scikit-learn, TensorFlow, and Librosa through coursework and hands-on experimentation.

Highlights

  • Implemented classic algorithms and data structures from scratch to build problem-solving fundamentals.
  • Mastered relational database design using PostgreSQL and SQLite through academic labs and application development.
  • Built proficiency in Git, GitHub, Linux environments, and collaborative software development workflows.

“Your love makes me strong, and your hate makes me unstoppable”

© 2026 Advaith R Pai. All rights reserved.