An AI researcher and backend engineer bridging machine learning models with production systems. They previously led engineering at VOGA through its acquisition by BTG Pactual, Latin America's largest investment bank, architecting a real-time data platform managing over $300M in assets on AWS. At The University of Tokyo, they developed recommender systems with Nikkei and Toyota, publishing an avoidance-aware recommender at SIAM SDM'25 and the open-source NewsReX framework at CIKM'26. They are now completing a Master of Research in AI and Machine Learning at Imperial College London, researching representation collapse in contrastive learning and mixture-of-experts models. They build scalable deep learning pipelines with PyTorch and JAX.

01

Education

Sep 2025 — Sep 2026London, UK

Imperial College London

Master of Research (MRes), Artificial Intelligence & Machine Learning

Research areas: Contrastive Representation Learning, Long-Tailed Learning, Mixture of Experts, and LLMs.

Thesis A Minority of One: Binary Imbalance beyond Long-Tail Learning, supervised by Professor Pedro Mediano. Researched multimodal representation collapse in contrastive learning and mixture-of-experts models under data scarcity, and investigated expert collapse in mixture-of-experts routing when minority classes are scarce.

Aug 2016 — Jun 2022Brasília, Brazil

University of Brasília (UnB)

BSc Electrical Engineering — First-Class Honours, GPA 4.35 / 5.0

Research areas: FPGA co-processors, machine learning for audio codecs, and deep learning for finance.

Undergraduate research on FPGA hardware acceleration of SHA-3 cryptography for low-power IoT devices, funded by PIBIC/CNPq and supervised by Professor Alexandre S. Nery. Final thesis applying machine learning to financial market prediction, supervised by Professor Edson Mintsu Hung.

Jan 2017 — Dec 2018Brasília, Brazil

University Center of Brasília (CEUB)

Associate Degree, Systems Analysis & Development

Concentrated on Java development, particularly real-time tracking systems, with a final thesis on a pharmacy delivery application under the guidance of Professor Auto Tavares.

02

Experience

Apr 2023 — Apr 2025Tokyo, Japan

The University of Tokyo

Research Assistant · supervised by Prof. Toyotaro Suzumura and Dr. Yuichiro Yasui

Nikkei (Financial Times) collaboration, news recommendation Aug 2023 – Apr 2025

  • Built data pipelines and feature engineering over Nikkei user interaction logs to train news recommendation models.
  • Modelled user behaviour and implicit negative feedback (news avoidance) with candidate-aware attention to mitigate popularity bias, accepted at SIAM SDM 2025.
  • Built NewsReX, a dual-backend PyTorch/JAX framework for reproducible news recommendation with up to 2.8x faster training, accepted at CIKM 2026.
  • Built multi-agent LLM pipelines (Strands Agents) with agentic retrieval, local open-weight LLMs and retrieval-augmented generation to extract user behaviour patterns, evaluated with LLM-as-judge.

Toyota collaboration, mobility and point-of-interest recommendation Apr 2023 – Aug 2023

  • Processed GPS vehicle telemetry into trajectories and user–place graphs for next-destination prediction.
  • Implemented graph neural networks and the lab's published methods in PyTorch to model driver mobility.
  • Designed the benchmark against a range of baselines on Toyota's proprietary data and presented the findings.

Jul 2021 — Apr 2023Brasília, Brazil

VOGA, acquired by BTG Pactual

Intern → Developer → Development Team Lead

  • Architected a real-time portfolio data platform (Flask, Next.js, PostgreSQL) monitoring over USD 300M in client assets.
  • Built cloud infrastructure as code on AWS (ECS, RDS, EC2) with Terraform, Docker and Jenkins CI/CD.
  • Automated reporting workflows into ETL pipelines and self-service dashboards.
  • Led a team of 7 engineers through the post-acquisition integration into BTG Pactual, Latin America's largest investment bank.

Sep 2020 — Jul 2021London, UK (Remote)

Cellcrypt Inc. — Advanced Security Systems

Applied Machine Learning Intern · supervised by Prof. Edson Mintsu Hung

  • Built an ML training and inference pipeline on AWS EC2 that tunes VoIP codecs, improving call quality by 7%.
  • Analysed packet captures (Wireshark) and call logs to diagnose quality loss in encrypted calls.
  • Built a C++ VoIP testbed (PJSIP, Opus, G.711) for codec experimentation.
03

Research Papers

  1. A Look Into News Avoidance Through AWRS: An Avoidance-Aware Recommender System
    January - July 2024 | SDM'25 Proceedings | ArXiv
    • Collaboration: Toyotaro Suzumura (The University of Tokyo) and Yuichiro Yasui (Nikkei Inc.)
    • Highlights: Developed AWRS, an Avoidance-Aware Recommender System for news that incorporates article avoidance as a key factor to improve recommendations. Evaluated on datasets in English, Norwegian, and Japanese, AWRS outperformed existing methods by leveraging avoidance as an indicator of user preferences.
  2. NewsReX: An Open-Source Multi-Framework for Neural News Recommendation
    January - August 2025 | CIKM'26 | ArXiv
    • Collaboration: Toyotaro Suzumura (The University of Tokyo) and Yuichiro Yasui (Nikkei Inc.)
    • Highlights: An open-source framework that unifies multiple neural news recommendation models across two ML backends (PyTorch and JAX/Flax). Models are defined through a structured YAML specification that decouples model definition from framework-specific implementation, with a Hydra-based configuration system for composable experiments and single-command framework switching. Cross-framework parity tests verify numerical consistency across backends, and pre-trained weights are published so results can be reproduced by inference alone.
  3. POPK: Mitigating Popularity Bias via a Temporal-Counterfactual
    April - July 2024 | ArXiv
    • Collaboration: Toyotaro Suzumura (The University of Tokyo) and Yuichiro Yasui (Nikkei Inc.)
    • Highlights: Developed POPK, a model which uses temporal-counterfactual analysis to reduce popularity bias in news recommendations. POPK improves accuracy and diversity by systematically removing the influence of popular articles.
  4. A SHA-3 Co-Processor for IoT Applications
    January - November 2020 | Paper (IEEE - WCNSPS'20)
    • Collaboration: Alexandre S. Nery (University of Brasília) and Alexandre da C. Sena (Rio de Janeiro State University)
    • Highlights: Designed and implemented a SHA-3 hardware co-processor on FPGA for IoT applications, achieving 65% faster performance than ARM Cortex-A9 with improved energy efficiency and reduced circuit area.
04

Honours

  1. Japanese Government (MEXT) Research Scholarship April 2023 - April 2025 | About The Japanese Government (MEXT) Research Scholarship supports international students conducting research at Japanese higher education institutions. Approximately 9,000 students are accepted out of 100,000 global applicants (~9% acceptance rate), with full funding for tuition, living expenses, and research activities.
  2. Brazilian Government (CNPq) Institutional Scientific Initiation Scholarship (PIBIC) August 2019 - July 2020 | About The PIBIC program, funded by the Federal Government of Brazil, supports undergraduate students in research, technological development, and innovation. Awarded to the top ~4% of undergraduates nationwide on academic merit; funded research into FPGA-based cryptographic security systems.
05

Awards

  1. SIAM Travel Award – SDM25 May 2025 | About Granted by the Society for Industrial and Applied Mathematics (SIAM), this award supported travel to the 2025 SIAM International Conference on Data Mining (SDM25) held in Alexandria, VA. It recognized promising early-career researchers contributing to the field.
  2. Invited Conference Report – Database Society of Japan June 2025 | About Invited by Japan's national database society to write the SDM'25 report for its newsletter, 18(3).
  3. Y Combinator Startup School, Travel Grant July 2026 | About Selected as a high-potential builder for YC's technical conference in San Francisco, with talks by Sam Altman (OpenAI) and Jensen Huang (NVIDIA). Travel funded by YC.
06

Competitions

  1. 3rd Place – HackTheLaw, University of Cambridge June 2026 | About Hosted by the University of Cambridge and Stanford CodeX. Recognized for technical excellence and LLM integration by a panel drawn from academia, industry (Perplexity, Anthropic), and law firms (Clifford Chance and White & Case). Built Quinn, an AI platform that keeps lawyers across the facts of large, fast-moving cases.
07

Open Source

  1. learning-ai.cafe - Website
    A collaborative, open-source AI textbook with one tutorial per topic, guiding readers from first intuition to deep understanding, with runnable code exercises. Anyone can help clarify and improve it.
  2. NewsReX - Code | Weights
    An open-source framework for neural news recommendation, built with The University of Tokyo and Nikkei.

    It is research tooling: models are declared in YAML and composed through Hydra, and run on two backends — PyTorch and JAX/Flax — behind a single interface. Cross-framework parity tests verify that both implementations agree numerically before any result is reported, and evaluation is shared and framework-agnostic. Everything is MIT-licensed and every trained weight is public, so results are reproducible by inference alone.

    Shared by François Chollet, the creator of Keras.
  3. NewsrecLib - Code
    Implemented the PP-REC SOTA model into the news recommendation framework.
  4. Qlib - Code
    Added support for the Brazilian stock market, enabling local investors and researchers to use Qlib's machine learning models and data processing pipelines on Brazilian stock data.
08

Certifications

  • IELTS: Overall Band Score: 8.0 (Listening: 8.0, Reading: 8.5, Writing: 7.0, Speaking: 8.5)
09

Essentials

Courses

  • Deep Learning @ Carnegie Mellon University | Certificate
  • Digital Signal Processing @ EPFL | Certificate
  • Information Theory @ The Chinese University of Hong Kong | Certificate
  • Algorithms and Data Structures @ UCSD | Certificate

Leadership experiences

  • Business Development Lead, Electrical Engineering Junior Enterprise (ENETEC) June 2019 - April 2020 ENETEC is a student-run, non-profit consultancy at the University of Brasília. Closed about USD 15,000 in consulting contracts and contributed to ENETEC winning the High Growth, Connected and Impact Junior Company awards in 2019 from Brasil Júnior.
  • Development Team Lead, VOGA October 2022 - April 2023 Managed and mentored a team of 7 engineers across frontend, data and operations through the post-acquisition integration into BTG Pactual.

Languages

Portuguese Native
English Near-native
Spanish Conversational
Japanese Basic

Tech stack

Programming Languages

Python JavaScript SQL C++ C Java

Frameworks & Libraries

PyTorch Lightning AI JAX PyTorch Geometric scikit-learn TensorFlow Keras Plotly Strands Agents Flask FastAPI

Infrastructure

AWS Cloudflare Docker Nginx

Databases

PostgreSQL

Igor L.R. Azevedo

Master of Research · AI/ML
Igor L.R. Azevedo
Contrastive learning Recommenders LLMs Electrical Eng.
🇬🇧 🇯🇵 🇧🇷 igorazevedo.com