Daniel Lobato

explore the graph
The Bio, Flowing
move your mouse to disturb the field — text reflows in real time via canvas layout, no DOM
movement I — retrieval
Ask me anything

I work on search & retrieval at Instagram. So naturally, this site has its own retrieval engine. Type a query and watch the "embeddings" find the nearest match.

movement II — forward pass
Career Graph
2021 — present
Machine Learning Engineer
Meta — Applied Research on MRS, Monetization, Instagram
I've worked at Meta for the last few years on a variety of teams. Right now, I am working mostly on Instagram Search, specifically on Retrieval and Query Understanding problems. We are trying to make our platform the best content search in industry. Before then, I bootstrapped the monetization efforts for conversion and user quality models in Marketing Messages on Whatsapp. My initial role in Meta was working at Meta AI, on Modern Recommendation Systems. I've worked on Graph Neural Networks (GNNs) at Facebook and Instagram scale for a variety of recommendation systems. I helped with some testing automation in Iron Man VR too, never thought I'd show up on a game credits!
GNNs RecSys Search & Retrieval Query Understanding
2021
Machine Learning Engineer
Palo Alto Insight
I worked for Palo Alto Insight, writing software to help deploy machine learning based solutions, and automate workflows with AI. We were offering consulting services mostly in data analysis and ML for vision to companies who needed it.
Computer Vision MLOps Consulting
2014 — 2021
Senior Software Engineer
Red Hat
I spent 7 years working at Red Hat, working on Red Hat Insights, a tool for continuous analysis of your applications to predict risk, recommend actions and track costs in a hybrid cloud setting. My first 4 years there, I worked on the Foreman project and Satellite, which were the best options out there for hybrid cloud+bare metal management.
Open Source Hybrid Cloud Red Hat Insights Foreman
2012 — 2014
Open Source Liaison
CERN — Geneva, Switzerland
I spent 2 years working at CERN acting as liason between different Open Source projects (Puppet, Foreman and others) and our lab needs.
Foreman Puppet Open Source
interlude — backpropagation
Education
🐝
M.S. Computer Science
Georgia Institute of Technology
Completed an MSCS with a specialization in Machine Learning.
Machine Learning OMSCS
🗽
hackNY Fellow, Class of 2012
hackNY
A transforming experience. I was a hackNY fellow and a mentor of the 2015 class. Here's my lightning talk from that summer.
NYC Startups Fellowship
🎓
B.S. Computer Science & Engineering
Universidad Carlos III de Madrid + UC Irvine
Got my degree from UC3M and studied 1.5 years at UC Irvine. Did my last year completely remote from Geneva, Switzerland, as I was working from CERN at that time, and flew back only for exams.
Madrid California CERN Remote
📚
Publications
Google Scholar
Part of me wishes to have joined a PhD program, but it just never happened. However, I'm happy to collaborate on publications that align with my interests.
Research Collaboration
movement III — embeddings
Places I've Called Home

Seven cities, four countries, two continents. Each one a node in the graph, each edge a flight I'll never forget.

🇪🇸
Madrid
Spain
Born and raised here. Grew up in Leganés, then lived in Barrio de las Letras. Beautiful strolls around Malasaña and nights at Next Level Bar. I studied Computer Science and Engineering at Universidad Carlos III de Madrid.
🇺🇸
Irvine
California, USA
I studied 1.5y at University of California at Irvine. Fondest memories: late night BBQs at Corona del Mar and long runs all the way to Laguna Beach.
🇨🇭
Geneva
Switzerland
I did my last year of uni completely remote from Geneva while working at CERN, flying back only for exams. Fun fact: I regularly crossed the French/Swiss border just walking through the CERN cafeteria!
🇮🇱
Ra'anana
Israel
Near Tel Aviv. I loved the falafel and taking culture trips around the area — especially Herzliya and the Old City of Jerusalem.
🇬🇧
London
United Kingdom
Where I met my now wife. I miss biking around the city and sipping flat whites and cocktails. Best summers in the western hemisphere.
🇺🇸
New York City
New York, USA
I was a hackNY fellow, class of 2012, and a mentor of the 2015 class. Transforming experience. Not the biggest fan of the dirtiness, but I love kayaking on the river and wandering SoHo / East & West Village. Find me south of Madison Square Park, wherever there aren't too many skyscrapers.
🇺🇸
Jersey City
New Jersey, USA
Where I live with my wife and son, close to beautiful Van Vorst Park, across the Hudson river from Manhattan. Love Liberty State Park, the waterfront, one of the best roasters in the country (Modcup), the best pizza (Razza), and an incredible set of Chinese restaurants.
movement IV — latent space
Beyond the Loss Function
🎹
Piano
I like to play the piano, at about ~ABRSM 8 level, working on DipABRSM. I'm interested in chamber music collaborations, violin duos, so if you're into that hit me up!
🕹️
Dance Dance Revolution
It's the one thing that brought the most relationships into my life, made me met a lot of people, and I really enjoyed. I ranked #2 in Spain consistently for many years, but it requires a time investment I don't want to make right now.
🏔️
Skiing
I'm always down to hit the slopes once ski season has started. My favorite resorts are Zermatt, Grindelwald and Grandvalira.
🇨🇳
Mandarin Chinese
I'm currently learning Mandarin Chinese. My main motivation is being able to communicate with my partner's family and friends. Actually, learning Japanese (~N4) has helped me tremendously with learning Chinese characters.
💪
Calisthenics & Boxing
I enjoy calisthenics to keep my body in shape. I recently hit a plateau there so I moved to a personal trainer. I used to spar as a light-welterweight boxer. I thoroughly enjoyed it and hope some day I can get back to it.
👐
Open Source
I was one of the main maintainers of Foreman (2.3k+ stars) for 6-7 years. Grew the community across FOSDEM, Linux NYC, PuppetConf, AnsibleConf. I also publish academic paper notes on a public repository — a mix of OS and ML papers from my Masters onward.
movement V — training data
Notes
🍇
Graph Neural Networks: Vanilla, GCN, GAT, GIN
Machine Learning
read →
🔗
Graph Neural Networks for Recommendation Systems: Link Prediction
Machine Learning
read →
🧬
GPT-GNN: Generative Pre-Training of Graph Neural Networks
Machine Learning
read →
📌
PinSage: Graph Convolutional Neural Networks for Web-Scale Recommender Systems
Machine Learning
read →
👁️
How attentive are GATs? (GAT v2)
Machine Learning
read →
💎
How Powerful Are Graph Neural Networks (GIN)
Machine Learning
read →
#️⃣
HashGNN: min-hashing & locality sensitive
Machine Learning
read →
🌐
DistDGL: Distributed Graph Neural Network Training for billion-scale graphs
Machine Learning
read →
🤖
Do Transformers Really Perform Bad for Graph Representation?
Machine Learning
read →
A Generalization of Transformer Networks to Graphs
Machine Learning
read →
🤝
Neural Graph Collaborative Filtering
Machine Learning
read →
🦭
Link Prediction based on Social Networks — SEAL
Machine Learning
read →
📊
DLRM — Deep Learning Recommendation Model
Machine Learning
read →
✖️
DCN V2: Improved Deep & Cross Network
Machine Learning
read →
📎
CLIP: Contrastive Learning Image Pretraining
Machine Learning
read →
🏭
AGL: A Scalable System for Industrial-purpose Graph Machine Learning
Machine Learning
read →
🎯
Graph Attention Network
Machine Learning
read →
🖼️
AlexNet
Machine Learning
read →
📏
Evaluation tasks for LLMs: Perplexity, BPW, log nats…
Machine Learning
read →
🌗
Infinite bits and the rule of product
Technology
read →
💎
Code Tuning: a programming pearl in Ruby
Technology
read →
🔄
Corecursion and Codata
Technology
read →
🐻
29 things that make you a cool person
Introspection
read →
🧘
Personal Capital
Introspection
read →
coda
Play something

Every node in the graph above plays a note. But if you'd rather play yourself:

click the keys or use keyboard: A S D F G H J K (white) & W E T Y U (black)

Repertoire

~ABRSM 8 level, working on DipABRSM. Interested in chamber music collaborations and violin duos.

Baroque

Bach2-Part Invention No. 8 in Frecordedvideo
BachSinfonia No. 8 in F Majorforgotten
BachSinfonia No. 9 in F minorrecordedvideo
ScarlattiSonata 366 K1 D minorrecordedvideo

Classical

BeethovenPiano Sonata No. 1, Op. 2 No. 1recordedvideo
Beethoven9 Variations on Paisiello WoO 69forgotten
BeethovenSonate No. 8 "Pathétique" — 2nd Mvtrecordedvideo
HaydnSonata Hob. XVI:5 — I. Allegrorecordedvideo
HaydnSonata Hob. XVI:5 — II. Menuetrecordedvideo
HaydnSonata Hob. XVI:5 — III. Prestorecordedvideo
MozartSonata in A minorrecordedvideo
MozartFantasia in D minorrecordedvideo
MozartPiano Sonata No. 16, K. 545recordedvideo

Romantic

ChopinNocturne in C-sharp minor, B.49recordedvideo
ChopinEtude Op. 25, No. 1 "Aeolian Harp"reading
MendelssohnLieder ohne Worte, Op. 53: Andante con Motorecordedvideo
SchubertImpromptu No. 2recordedvideo
SchubertImpromptu No. 3recordedvideo

XX Century

AlbénizEspaña, Op 165: 1. Preludiorecordedvideo
BartókRomanian Folk Dances: III. Pe-Locrecordedvideo
DebussyClair de Lunerecordedvideo
DebussyGollywog's Cakewalkrecordedvideo
ProkofievVisions Fugitives: II Andanteforgotten
ProkofievVisions Fugitives: X Ridicolosamenteforgotten

Etude

SibeliusEtude Op 76 No. 2abandoned

The Five Zones of Practice

A framework I follow for structuring practice sessions.

New Material
  • Divide into sections
  • Establish interpretive/technical plan
  • Slow tempo
Developing Material
  • Refine interpretation
  • Increase tempo
  • Memorize
Performance Material
  • Practice performing
  • Maintain memory
  • Renew and innovate
Technique
  • Scales
  • Arpeggios
  • Voice/instrument-specific work
Musicianship
  • Sight-reading
  • Theory/ear training
  • Composition/improvisation
  • Listening/study