Profile banner
Chirag Ahuja

Chirag Ahuja

Senior applied scientist in ML/NLP; PhD in signal processing

ahujachirag
Bengaluru, India
Joined March 2026
Favicon imagechiragahuja.com
Favicon imagegithub.com
Favicon imageamazon.science
Favicon imagescholar.google.com.sg
Favicon imageresearchgate.net

Network

1.5K connections
🤖
AI/ML Innovators
🤝
Talent Acquisition Leaders
⚙️
Product Engineering Leaders
🧑‍💻
HR People Operations
🚀
Startup Founders
🔬
...
🎓
EdTech Upskilling

Summary

Applied industrial scientist with multi‑company experience: Chirag has held senior applied-science and machine-learning roles across industry (Tatras/Tatras Data, Visteon, Oracle, ServiceNow, Amazon), applying ML and deep-learning methods to product problems ranging from automotive ADAS/voice systems to forecasting and advertising trust. theorg+2
Researcher in physiological signal processing and EEG: He pursued a PhD in Physiological Signal Processing at Delhi Technological University and has authored publications and projects focused on EEG emotion recognition and signal‑processing methods. ac+2
Open-source practitioner and educator: maintains a personal website with technical tutorials and a GitHub with reproducible research code (notably TRANSIT-EEG), indicating an emphasis on sharing methods, code and practical guides for deep learning and signal processing. chiragahuja+1
Inventor and systems thinker in time-series and forecasting: listed as inventor on multiple patent filings (Oracle-assigned) covering forecasting, hyperparameter tuning and time-series processing, reflecting applied research contributions in forecasting and decision systems. justia+1

Work

Education

Projects

Writing

US20240386047A1 - Cold-start forecasting via backcasting and composite embedding

November 1, 2024

Patent application describing techniques for cold-start forecasting using backcasting and composite embeddings to combine metadata text embeddings with time-series vectors for forecasting in absence of direct historical data.

Favicon imagepatents.google.com

ADAPTER: Auto-Augmentation for Emotion Recognition in EEG - A Class and Subject Invariant Approach

Conference paper introducing ADAPTER, an automatic data-augmentation strategy for EEG-based emotion recognition that finds class- and subject-invariant augmentation policies and reports improved accuracy on SEED and related datasets.

Favicon imageieeexplore.ieee.org