Bangalore, India

New — MAZEBORN: Relic Hunt is out on Google Play

Adithya Tadaga
Gangadhar

Data Scientist at Target · Search & Ranking

I build the relevance models and multi-armed bandits that decide what millions of Target shoppers see when they search.

View my work Read my research

01 · About

Search, ranking, and the systems behind them.

I work on information retrieval at Target — contextual relevance models, multi-armed bandits for listing pages, and type-ahead search ranking. Outside work I ship my own products, most recently MAZEBORN: Relic Hunt, a 3D maze game on Google Play. Earlier I built facial-recognition and text-to-speech systems in production, and published 11 peer-reviewed papers across NLP, information retrieval, computer vision and healthcare AI, presented in Sweden, Italy, Japan and India. Current focus: making product discovery feel less like querying a database and more like asking someone who knows the store.

Adithya at Canary Wharf, London

02 · Experience

Where I've worked

  1. Aug 2025 — Present

    Data Scientist · Target

    • Building dynamic listing-page experiences using contextual relevance models and multi-armed bandits to improve search relevance and product discovery.
    • Improved type-ahead search ranking and suggestion quality across the search experience.
  2. Feb 2025 — Aug 2025

    AI Data Scientist Apprentice · Target

    • Developed search relevance, retrieval, and ranking models using DeBERTa, LLMs, embeddings, and behavioral interaction signals.
    • Built automated ML pipelines and experimentation workflows on GCP with Vertex AI, BigQuery, and Python.
  3. May 2023 — Feb 2025

    AI Engineer · Maacsys

    • Enhanced facial-recognition workflows for automated access-control systems, improving recognition accuracy by 7%.
    • Built a custom text-to-speech application using deep learning models and Python automation workflows.
  4. Jun 2024 — Jan 2025

    Teaching Assistant, Deep Learning · PES University

    • Taught Transformer architectures, attention mechanisms, and modern deep learning techniques to postgraduate students.
  5. Dec 2023 — Jul 2024

    Research Intern · PES Center for Data Science & AIML

    • Published research on comparative analysis and evaluation of fine-tuned large language models, contributing to multiple peer-reviewed publications.

03 · Projects

Selected work

Live XAUUSD Forecasting System

Live in production

Problem

Gold (XAUUSD) trades 23 hours a day and reacts to the dollar index and US 10-year yields within minutes. Manually tracking those cross-asset relationships on 30-minute bars isn't feasible.

Approach

Time-series models trained on engineered features across XAUUSD, DXY and US10Y — market structure, liquidity, volume profile and cross-asset correlation — with walk-forward validation to avoid look-ahead bias.

What I built

The entire system: data ingestion from live feeds, the feature pipeline, gradient-boosted models with automated retraining, signal generation with confidence gating, and trade logging with cooldown logic.

Outcome

Running live and unattended, generating ≈ $27.5/day at modal performance with full monitoring of every model decision.

  • Python
  • CatBoost
  • pandas
  • scikit-learn
  • FastAPI

MAZEBORN: Relic Hunt

Live on Google Play

Problem

Most mobile maze games are static — once you've memorized the layout, the tension is gone. I wanted a maze that stays threatening on the fiftieth run.

Approach

A first-person 3D labyrinth that changes between runs, with patrolling Watcher enemies supplying the pressure instead of countdown timers. Collect four Golden Relics, unseal the Master Relic, and escape through the portal.

What I built

The whole game, solo — maze environments, first-person controls, Watcher enemy behavior, the relic progression loop — through to release on the Play Store.

Outcome

Published on Google Play as a first-person adventure game, live and installable today.

Vision Models for Text-Heavy Content

Research · arXiv

Problem

Vision models handle natural images well but struggle with documents, charts and dense text — content that dominates real-world use.

Approach

Evaluated and enhanced vision-language models for understanding and interacting with text-heavy visual content.

What I built

Co-authored the study and its evaluation methodology; published as a preprint with four co-authors from PES University.

Outcome

Published on arXiv (May 2024); part of a research track that led to 11 peer-reviewed publications.

04 · Publications

Research

Research spanning NLP, information retrieval, computer vision, healthcare AI and human-computer interaction — presented in Sweden, Italy, Japan, Bahrain and India.

  1. 2025

    GEN AI-Driven Automation for Wrist X-ray Report Generation

    CIBB 2025 — 20th International Conference on Computational Intelligence Methods for Bioinformatics and Biostatistics, Milan, Italy

    Conference
  2. 2025

    Leveraging Virtual Reality and AI Tutoring for Language Learning: A Virtual Campus with GPT Integration in Unity 3D

    HCI International 2025 — 27th International Conference on Human-Computer Interaction, Gothenburg, Sweden

    Conference
  3. 2025

    Comparative Analysis of Computer Vision Models for Image Classification on Wrist X-rays in the MURA Dataset

    Grenze Journal of Engineering and Technology

    Journal
  4. 2024

    Comparative Analysis of Large Language Models Fine-tuned on a Singular Dataset

    NLPIR 2024 — 8th International Conference on Natural Language Processing and Information Retrieval, Japan

    Conference
  5. 2024

    Automated Flying Drone Deployments for Search, Rescue and Probing Missions Using Digital Image Processing

    ICETAS 2024 — IEEE 9th International Conference on Engineering, Technologies & Applied Sciences, Bahrain

    Conference
  6. 2024

    Optimization of Cornering Lights in Adaptive Headlamps

    INCET 2024 — 5th International Conference for Emerging Technology, India

    Conference
Google Scholar profile →

05 · Education

Education & honors

PES University, Bangalore

B.Tech in Computer Science & Engineering

CGPA 8.66 / 10

  • CNR Rao Award for academic excellence
  • Academic Excellence Scholarship and multiple Distinction Awards
  • Winner, Embrione 2.0 Hackathon
  • 700+ day streak on LeetCode

06 · Skills

Tools I work with

Machine Learning & AI

  • Search Ranking
  • Information Retrieval
  • Recommendation Systems
  • NLP
  • LLMs
  • RAG
  • Semantic Search
  • Deep Learning

Programming

  • Python
  • SQL
  • Java
  • FastAPI

Cloud & Data

  • GCP
  • Vertex AI
  • BigQuery
  • Docker

Languages

  • English
  • Kannada
  • Hindi
  • Swedish (A2)
  • French (basic)

07 · Contact

Say hello.

The fastest way to reach me is email — I read everything.

adithyatg.work@gmail.com LinkedIn GitHub