– Gaia is my demanding muse
As told by Vasily Belokurov

First memory
What is your earliest memory? The kind that returns to you in surprisingly clear and vivid detail? My first “astronomical memory” is this. I am in an empty schoolyard, under a gloomy, clouded sky, surrounded by tundra, some 6,000 km from home, a loudspeaker is blasting the ominously glorious “Welcome to the Machine” by Pink Floyd and, indeed, the others in our small group and I are attending to our “machines”: we have telescopes poised to observe a total solar eclipse. The date is 22 of July 1990, I am 14 years old, and I am out of luck.

How did I end up here? I got bored and restless. I got bored because I ran out of sci-fi books. I got restless because I failed to join the first Soviet computer club, opened by the then world chess champion Garry Kasparov.

The past few months of preparation, the long expedition to get here lugging the numerous boxes of gear, feeding the hungry local mosquitoes and vicious midges with my young blood, all appear to have been in vain. The Sun and its magnificent corona will stay hidden behind the clouds... But just in time, a hole opens in the heavens, revealing a small patch of blue and we greet it with screams of excitement and relief. I glance at the three pestle-shaped cartridges in front of me: these are the cassettes I loaded with film just minutes earlier in the school bathroom. I need to focus now, our ambitious plan, set out by my always smiling advisor Anna Delaunay, is to make several exposures during the two minutes of totality, the two minutes the whole past year has been building toward.

New school
How did I end up here, in the middle of the tundra, staring in awe at this cosmic spectacle and manically gesticulating to get the midges off my face? The short answer is that I got bored and restless. I got bored because I ran out of sci-fi books at home (Arthur C. Clarke, Ray Bradbury, Stanislaw Lem). I got restless because I tried and failed to join the first Soviet computer club, opened by the then world chess champion Garry Kasparov.

Bored and restless
I got bored and restless enough to petition my mum, Tatiana, to let me change schools. To her great credit, she listened, trusted me, and helped me act on it. Tatiana suggested a specialist maths and physics school, and took me there for an interview. As I remember it, most interview questions left me stumped, but I got in. My new school came with a long-ish commute, new friends, and a much closer connection between teachers and pupils. My two maths teachers, Igor Varshavski and Marina Tsirlina, were unlike any I had known before. They quickly won us over with their informal attitude and passion. They also left us entranced at times by their oracular formulations of abstract mathematical concepts, which sounded alien and cryptic yet compelling. It also mattered that with them, I could both discuss the latest band I was into (Zvuki Mu) and try, somewhat audaciously, to earn an excellent final grade by beating one of them at chess (attempted many a time but never achieved).

Up on the roof
Up until then I had been growing up as a normal Soviet kid, dancing ballet (with my pal Anton) and collecting scrap metal with the other Young Pioneers, but the new school offered new ways of filling time. One day my friend Serge dragged me to a city-wide olympiad. He had a clear goal: to try the Astronomy section, and I went along because I had nothing better to do. We both tried to solve some astro puzzles and agreed that it was tricky but fun enough. Serge discovered that the organizers ran an evening astronomy school, which he insisted we absolutely had to join. This may have been the point of no return for me.

I can still remember long winter nights on this remote Uzbek mountain, when we took turns to observe Andromeda in search of variable stars.

The astronomy club simply had too many appealing features: we got to stay late, sometimes overnight (“yes, Mum, sorry, Mum, we are observing tonight; did you see the sky is clear?”); we had our own lair just under the roof of the evening school’s building; better still, we got to hang out on the roof at night; and we had a computer with “Prince of Persia” installed. In the club we learned the theory: the properties of planets and comets, the basics of spectroscopy and photometry, the types of telescopes. The main attraction, however, was the practical part: in small groups, we travelled to the principal observatories in the country and were given a chance to “drive” some of the big professional telescopes. Back then, to take a one-hour long exposure on Maidanak’s one-metre telescope, we had to guide by hand to correct for slight drift. I can still remember long winter nights on this remote Uzbek mountain, when we took turns to observe Andromeda in search of variable stars. Between exposures we would come into the observer’s hut from the cold for a quick cup of tea and a spin through an eclectic selection of LPs left behind by earlier visitors. The one we played a lot in the winter of 1991 was “Stabat Mater” by Pergolesi.

University
My first astronomy mentor, Natalia Kulakova, was there with us on many of these trips, from the Uzbek winter wilderness to the sun-bleached Crimea. With her calm, philosophical take on things, she seemed to herd us wild teenage cats with ease. She supported our chaotic ramblings and pushed us to explore more, to take part in these numerous school “expeditions”. She was also a “parent whisperer,” reassuring our worried mums and dads as we prepared to set off on each new wild adventure. Natalia encouraged Serge and me to try our hand at lecturing in a shiny new, almost magical gizmo: one of the first inflatable planetariums brought into the USSR from the UK. This was our first “proper” summer job. Even though we were each paid a modest salary, we were more than compensated by the gasps of awe from members of the public sitting on the floor in the dark with us. Kulakova also found a project supervisor for me when I entered the integrated Astronomy Master’s program at Moscow’s Lomonosov University in 1993, and with her help I could also begin a research project from day one with one of the professors at the Sternberg Astronomical Institute, Mikhail Sazhin. I had been dropped into the middle of one of the most active astrophysics departments in the country: the relativistic astrophysics laboratory headed by Nikolai Shakura. If you were into black holes, neutron stars, accretion disks, and the Cosmic Microwave Background, this was the place to be. It was a rather small department that punched above its weight, with a constant stream of visitors and several regular talk series, including a seminar on Gravitation and Cosmology.

Gravitational microlensing, big data and exotic matter
Yet my initial excitement about cosmology quickly subsided when Prof. Sazhin tried to instill in me a love of Christoffel symbols. Seeing my lack of enthusiasm, Sazhin changed tack and offered a different topic: gravitational microlensing. In the early 1990s, gravitational microlensing embodied the zeitgeist of observational astronomy: several ambitious sky surveys were beginning to monitor millions of stars night after night, creating some of the first truly large astronomical datasets. At the time, the community’s hopes were pinned on microlensing as a way to reveal the nature of Dark Matter. Dark Matter was the elephant in the room, except the room was the entire Universe and the elephant was holding the whole thing together while refusing to be seen. It was apparently everywhere, gravitationally dominant, and yet invisible. The mystery was not just how much of it there was, but what it actually was.

Astronomers felt Dark Matter’s presence acutely. There was plenty of evidence, from galaxy clusters (thanks to the pioneering work of Fritz Zwicky) and from the rotation curves of spirals (thanks to the pioneering work of Vera Rubin), for an additional, unaccounted-for mass component in the Universe. One possibility was that Dark Matter was not new at all, but consisted simply of compact, faint, or even non-luminous objects, such as black holes, neutron stars, and the less exotic red dwarfs, brown dwarfs, and old white dwarfs. The original idea of monitoring millions of stars in search of an extremely rare microlensing event was, I believe, due to Ken Freeman, who inspired Bohdan Paczynski to work out the details of such an experiment in his seminal 1986 paper. If we could observe and count all microlensing events, the contribution of ordinary stars to their total rate would tell us whether dark compact objects were responsible for Dark Matter. If they were not, then the answer was likely to be some kind of new particle yet to be discovered. These neutral, weakly interacting particles would underpin structure formation in the Universe and assemble into a hierarchy of scales. It became clear that detecting the smallest Dark Matter lumps possible would reveal the nature of the particle. Inspired by this idea, and by the ongoing microlensing campaigns, Alexander Gurevich proposed a way of detecting the smallest objects formed out of the “gas” of the hypothetical particle neutralino. The so-called neutralino stars should be amenable to gravitational microlensing observations, so Sazhin and I wrote a paper assessing the properties of those hypothetical events. This paper, born to the shadowy, menacing sound of Massive Attack’s album Mezzanine, was my entry into the world of large surveys and Dark Matter substructure.

Oxford
Off the back of that first small success, I decided to stick with gravitational lensing and microlensing, and try my luck with PhD applications abroad. So, in early 2000, I began scanning the web pages of different departments in Europe and found that the University of Oxford’s Theoretical Physics Department listed Dr Wyn Evans as someone with an interest in lensing. After a quick exchange of emails, Wyn pointed me to the Dulverton Scholarship, which provided financial support for students from Central and Eastern Europe. I applied and, to my shock, a few months later found a letter in my mailbox announcing that my application had been successful.

I managed to establish contact with a mysterious girl with cascading curls who would eventually become my wife. She taught me some Catalan and politely pretended to enjoy tea. She also pointed out that I was spending far too much time with my other muse – Gaia.

The first few months in the city of spires were spent rubbing my eyes in disbelief, as though the whole thing might vanish if I looked away. St Giles, where the Theoretical Physics Department then stood, and Novye Cheryomushki, the Moscow suburb I come from, looked as if they belonged on different planets, so I can be forgiven for feeling like Alice in Wonderland most of the time. In a place where everything looked curious and queer, music remained my coordinate system, the reference frame I could rely on. So one day, using Tom Waits and Lou Reed as beacons, I managed to establish contact with a mysterious girl with cascading curls who would eventually become my wife. She taught me some Catalan in exchange for some Russian, and politely pretended to enjoy the tea I naively made every time she dropped by. She also pointed out that I was spending far too much time with my other muse – Gaia.

Dreams of Gaia
Gaia, of course, was not a mistress, but a planned ESA space mission, a future observatory that promised to measure the positions and motions of stars with unprecedented precision. My introduction to it came through the first problem Wyn and I chose for my doctorate. Wyn offered me several starter themes. After a week or so of deliberation, we settled on a different flavour of microlensing: the astrometric kind. This idea was so far ahead of its time that it sounded like sci-fi. Astrometry offered a way of using microlensing not just to detect compact objects, but to measure their masses, by tracking the tiny wobble they induced in the apparent position of a background star. The catch was that this required a level of accuracy that only a space mission like Gaia could hope to reach. Many years after the end of my doctorate, Gaia would indeed begin detecting such minuscule wobbles in distant stars.

Just like Alice in Wonderland, I was undergoing rapid bodily transformations, doing my best to turn into a human-sized sponge: there was so much technical detail to absorb. With its unprecedented technological dexterity, Gaia was clearly envisaged as a jewel in humanity’s crown, yet I was in a panic, drowning in a sea of information. The helping hand that pulled me out was Erik Høg, the founding father of space astrometry, the mind behind Hipparcos -Gaia’s predecessor-, and the creator of the early Gaia concept. I met Erik at a conference, and during a coffee chat he quickly pointed out what was wrong with my understanding of Gaia. I made sure we stayed in touch.

Pondering astrometry was stimulating but, at the time, it still felt rather futuristic, so Wyn and I focused on a more immediate and pertinent problem for the remainder of the doctorate: with the current trawl of detected microlensing events, was there actually any room for dark compact objects? To answer this question, we decided to reclassify the measured light curves, which were available online for anyone to download in bulk. Observational astronomy was undergoing a paradigm shift: the early microlensing survey projects helped normalize the idea that a major astronomical experiment should not just publish papers, but also expose its underlying data products to the community through large public online archives. A postdoc working with Wyn, Yann Le Du, explained with great charisma that the only way we could pull off such a reclassification quickly and at scale was to use machine learning. After all, scientists at CERN, the European Organization for Nuclear Research already used neural networks to sift through millions of particle collision events, and so should we. Yann also announced that there was no chance of training a neural net on the computer I had been given and that we would have to build our own. And so we did.

Cambridge
Time flies when you are having fun. In 2003, with my thesis finally behind me, I arrived in Cambridge for my first postdoctoral position, carrying with me the machine Yann and I had built, a bundle of papers on Gaia, microlensing, and neural nets, unwavering support from Wyn, and a reference letter from the then PI of Gaia, Michael Perryman. The Institute of Astronomy in Cambridge seemed like the obvious next step: it was home to a renowned Hipparcos/Gaia team with Floor van Leeuwen and Gerry Gilmore at the helm. Immediately, I struck gold: I was given a single-occupancy office in one of the two appendices to the old Observatory building. My little room was too small for two people, but it was an absolute luxury: a space to concentrate, stay up late, or play “Little Room” by the White Stripes on full blast.

Compared to the dreamy, pensive atmosphere of Oxford’s Theoretical Physics, the IoA seemed like a whirlpool, a hive of intellectual energy, a gathering of minds. To get drawn in, one only had to show up for morning coffee in the foyer of the Hoyle Building. For me, the transformative event was the conference organized by Neil Trentham at the IoA in July 2005 to commemorate Donald Lynden-Bell’s seventieth birthday. Aptly named “The Mass and the Mystery in the Local Group,” the conference focused on a number of striking puzzles in our “own backyard.” The speakers shared the first clear hints of the “ghostly streams” predicted by Lynden-Bell, pointed to the excess of dark mass in our Galaxy’s halo, and debated the vexing “missing satellite” problem. Something was crystal clear: the paradigm shift that had begun with the Hubble archive and had been advanced by the first microlensing surveys was now in full swing. The Sloan Digital Sky Survey did not merely put the immense quantities of data it was collecting online, the collaboration set up SkyServer, which allowed astronomers worldwide to browse and query the database directly. This was a game changer. When the conference ended, I went, electrified, to my little “cave” to download the Sloan data and poke around.

The Field of Streams
My co-conspirator in this first Sloan exploration was Dan Zucker, who had previously used its data to study Andromeda’s outskirts. Together we decided to build as large a map of the Milky Way’s halo as the available data allowed. We were motivated by the recent predictions of Kathryn Johnston and James Bullock, who used computer simulations to show that stellar halos of galaxies like our own must be littered with tidal debris from disrupted dwarf satellites. If discovered, this messy, lumpy, entangled state of our halo would be both a natural consequence and a strong confirmation of the hierarchical galaxy formation paradigm built on Cold Dark Matter. Glimpses of this ongoing dwarf-galaxy merging had already been reported by teams led by Rodrigo Ibata, who discovered the Sagittarius dwarf right “under our noses”, and by Steve Majewski, who demonstrated that Sagittarius is rapidly falling apart.

Annoyingly, our machines were limited both in memory and in computing power, so we came up with a trick: first scrape Sloan’s SkyServer for every last byte of data, then split it into chunks, manipulate them to extract only the stars we were after, and finally stitch the map back together. The survey’s main goal was to map the structure of the Universe using galaxies as tracers, so the stars cluttering Sloan’s photographs of the sky were a nuisance. In fact, for us too, most stars were a nuisance, because Dan and I were after rare, unusual stars, the ones that were much older and much less chemically evolved than the rest. In the Sloan data, these oldies, born outside the Milky Way, were outnumbered by stars formed in our Galaxy’s disk by a factor of a thousand to one.

A picture is worth a thousand words. I know that in the age of AI this saying may not mean quite the same thing anymore, but twenty years ago it worked exactly like that.

Motivated by the earlier experiments with neural nets, I started with a model that predicted a star’s temperature and metallicity from its colours, which worked surprisingly well thanks to Sloan’s famous set of five filters: violet, blue-green, red, deep red, and near-infrared. But it quickly became apparent that, for our map-making, we could do away with the sophisticated machine-learning approach and select stars simply on the basis of a single colour and a single magnitude. Once we figured out the data handling and filtering, it was super easy to brush away thick layers of nearby, more metal-rich disk stars and reveal, hidden underneath, a panorama of stellar trails and clouds of all sizes. Dan came up with a brilliant name for the map we made: the Field of Streams.

A picture is worth a thousand words. I know that in the age of AI this saying may not mean quite the same thing anymore, but twenty years ago it worked exactly like that. Anyone who looked at the Field of Streams in 2006 and saw this colourful, glorious mess would go away thinking, “Wow, it is raining small galaxies!” This was widely seen as a big win for the structure formation theory based on cold dark matter. One of its main predictions, the so-called hierarchical assembly, meant that the Universe starts by creating small Dark Matter clumps that then merge to make bigger ones. With this paradigm in mind, the Field of Streams could be viewed as a kind of construction site: trails and clouds of stars are fragments of small building blocks being incorporated into our Galaxy’s halo, if only haphazardly. What’s more, it could be used as a time machine. Because the motions of stars follow predictable rules, we can work backwards in time and trace stellar streams to find out when their parent galaxies first entered the Milky Way. And the best part is this: as you trace the streams and model their paths, you get to weigh the dark matter they are travelling through. That is, if you can build the machinery to model streams. So the Field of Streams was a proof of hierarchical assembly, a map of streams, and a promise of a powerful new way to measure dark matter.

It’s a riddle wrapped in a mystery inside an enigma, Wyn once said, repurposing Churchill’s words to describe our lack of understanding

Galactic dance
Crossing the Field of Streams from end to end is the Sagittarius stream: the longest and brightest stellar stream in the Milky Way halo, and the first one to be mapped across the sky in detail. It is the debris of a small galaxy being torn apart by the Milky Way, its stars stretched into a vast ribbon that wraps around our Galaxy. In principle, this makes Sagittarius one of the best tools we have for weighing the Milky Way’s dark matter halo. The stream covers such a large volume that its path should tell us not just how much dark matter surrounds the Galaxy, but also how that dark matter is distributed.

There is only one problem: Sagittarius has stubbornly refused to make sense. To this day, we still have no complete model of its formation. “It’s a riddle wrapped in a mystery inside an enigma” – Wyn once said, repurposing Churchill’s words to describe our lack of understanding. For two decades, several groups have tried to build a model that could reconcile the many measurements of the stream. All in vain.

Then Gaia (what else, of course!) supplied a crucial missing piece. It showed, with startling clarity, that the Milky Way is not the quiet, balanced system we had often assumed it to be. Our Galaxy is out of equilibrium, it is being pulled, shaken, and distorted by its most massive companion, the Large Magellanic Cloud. Gurtina Besla and Nitya Kallivayalil were among the first to challenge the standard picture by showing that the Cloud was much more massive than previously thought. Once that was understood, the old problem began to look different. Perhaps Sagittarius was not misbehaving because we had misunderstood the stream alone; perhaps we had misunderstood the stage on which the stream was moving.

This is where stellar streams become so powerful. They are not just pretty trails of stars. They are sensitive tracers of gravity, recording the changing pull of the Milky Way, the Large Magellanic Cloud, and their dark matter halos. With Denis Erkal and Eugene Vasiliev, we have shown that many of the Milky Way’s satellites and streams are being tugged by the Large Magellanic Cloud. Sagittarius appears to feel this galactic dance most dramatically: not only through the Cloud’s direct pull, but also through the accelerated motion of the Milky Way itself. As the two galaxies swerve through their tango around one another, their dark matter halos are stretched, twisted, and displaced, breaking many of our earlier, simpler assumptions.

“Look at these dots!”
That was what Dan exclaimed, pointing at a surprisingly bright pixel in the Field of Streams. The image was peppered with these conspicuous, isolated “hot” pixels. But we did not think they were glitches. The Field of Streams was not really a photograph; it was a density map, so a bright pixel meant that the density of stars in that patch of sky was unusually high. We soon realised that many of these dots were small satellites of the Milky Way: compact clumps of stars orbiting our Galaxy.

Connecting the dots
The particular dot Dan was pointing at, however, was not listed in any catalogue. As far as we knew, it was unknown to anyone beyond our office – in other words, unknown to the rest of humanity. Was there reason to be excited? Absolutely. Until recently, the Milky Way had been known to have only about a dozen satellites, only a few more than the number of planets orbiting the Sun. Finding a new dwarf galaxy was a big deal in itself, but it mattered even more because theory predicted that many more should exist. Where were they? This was the so-called missing satellites problem, widely seen as one of the last major challenges to the otherwise spectacularly successful cold dark matter theory.

A simple solution now seemed to be staring us in the face: the satellites were not missing at all. They were just extremely well hidden. Our newly discovered dwarfs were far fainter than anyone had expected astonishingly faint – and it –took the Sloan Digital Sky Survey to begin revealing them. Beth Willman had already provided an early glimpse of this hidden population by identifying the first of these objects in the SDSS data a year earlier. Marla Geha and Josh Simon gave them the name that stuck: ultra-faint dwarfs.

We quickly automated the search, and it produced a remarkably long list of promising candidates to follow up with deeper, better data. That effort would not have been possible without Mike Irwin, a guru of astronomical data analysis. Looking over the results, Mike half-jokingly remarked that we were not being nearly adventurous enough, since our search kept turning up genuine dwarf galaxies and hardly any duds. Buoyed by this avalanche of discoveries, Dan suggested that we should call our little group the “satellite factory”.

In 2007, Sergey Koposov, a young PhD student, joined our group and transformed the community’s understanding of what the SDSS discoveries really meant. Sergey turned the raw tally of newly found dwarfs into the first estimate of the total number orbiting the Milky Way. He showed that if Sloan had uncovered only a handful, there must be hundreds still lurking below its detection limits. These were, unmistakably, the missing satellites predicted by cold dark matter theory.

The event that shook the Galaxy
It would be incorrect to say that Sloan’s view of the Galactic stellar halo perfectly matched theoretical expectations. The first clear sign that something was odd about the Milky Way’s halo came from Alis Deason, then a first-year PhD student at Cambridge. Just months into her PhD, Alis built a new model of the halo’s density using SDSS data. She did not just measure its 3D shape; she also asked how ragged, clumpy, or smooth it was.

Seventeen years after I first heard about Gaia, its first Data Release was finally out. It was no miracle, but the product of the dexterity, industry, and many sleepless nights of the Gaia Consortium

The answer was surprising. The halo appeared squashed, and its density dropped sharply beyond a certain distance from the Galactic centre - what Alis called a “broken” profile. Even more intriguingly, the halo was not as messy as expected. In fact, it looked smoother than many theoretical models predicted. That combination - a sharp break and unexpected smoothness - was the clue. In simulations, Alis found, such halos were not built from a constant drizzle of many small galaxies. They were the aftermath of something much more dramatic: a single ancient encounter with a massive dwarf galaxy. The Milky Way’s halo, in other words, looked less like a rubbish heap assembled from countless scraps and more like the wreckage of one major collision. This prediction remained waiting in the dark until Gaia arrived.

Seventeen years after I first heard about Gaia as a student, its first Data Release was finally out. It was no miracle, but the product of the dexterity, industry, and many sleepless nights of the Gaia Consortium, under the vision and guidance of its leader, Anthony Brown. In Cambridge, our impatient crew, helped by a neat technological masterstroke from Koposov, built an enhanced version of the first data release to peer into the inner workings of the Galaxy’s halo.

Gaia gave us something we had never had before at this scale: motion. It did not just show us where halo stars were; it showed us how they were moving. The first obvious sanity check was to ask whether stars with different amounts of heavy elements also had different kinds of orbits. But something very odd appeared in our plots. Instead of a broad zoo of orbital shapes, the bulk of the halo stars were moving on unmistakably radial paths, plunging in and out of the Galaxy like needles piercing its centre.

We called this striking, elongated pattern in stellar motion the Gaia Sausage; Amina Helmi’s group, tracing the same ancient event, named its progenitor Enceladus. It was as if we were watching the replay of a head-on collision: a dwarf galaxy had crashed into the young Milky Way, been torn apart, and left its stars flying back and forth through the Galactic centre. Could several smaller galaxies have conspired to fall in together, all on almost the same kind of orbit? That seemed possible in principle, but the stars themselves carried the answer. Their chemistry told us how massive their parent galaxy must have been before it was destroyed. Given how metal-rich the stars on these radial orbits were, the culprit looked very likely to be a single massive object. The Gaia data appeared to agree strikingly well with the Deason hypothesis.

Of course, hints of this ancient cataclysm had been seen before Gaia, and even before Sloan, but the evidence had been scattered and fragmentary. The discovery of Gaia Sausage/Enceladus changed that, rather as the discovery of the Chicxulub crater - the asteroid impact scar linked to the extinction of the dinosaurs - changed the debate about Earth’s past. The clues had been there for years, but now there was a single piece of evidence that was hard to argue away. The Milky Way, it seemed, had its own impact crater – not a hole in the ground, but a vast fossil pattern written in the motions and chemistry of its stars.

Gaia gave us something we had never had before at this scale: motion. It did not just show us where halo stars were; it showed us how they were moving.

And, like Chicxulub, this was not just an impact. It was a metamorphosis. The Gaia Sausage/Enceladus collision unleashed a chain of transformations across the Milky Way. The old Galactic disk was splashed and heated. The Galaxy’s pulse – its star formation – appears to have been disrupted. The new, younger disk in which our Sun was eventually born emerged soon after the collision, and so did the Galactic bar. The shape and orientation of the Milky Way’s dark matter halo were likely reset. Even the present-day system of dwarf satellites around the Milky Way may still carry the imprint of this ancient encounter.

This is why the event matters. If we want to understand galaxy formation and the behaviour of dark matter on the smallest scales, we first need to understand our own Galaxy as a historical object: not a static spiral in a textbook, but the survivor of a violent past. The Gaia Sausage/Enceladus event is one of the key chapters in that history. It tells us when the Milky Way was shaken, how it rebuilt itself, and where to look for the clues we have only just begun to gather.

Fortunate and grateful
It is, I hope, very clear from the short story above that none of the breakthroughs in my research would have been possible without the colleagues who generously shared with me their time, their energy and their ideas. Not only the great names in the field that I was lucky to cross paths with. I am indebted to the students and postdoctoral fellows who I nominally supervised, but in reality, learned a great deal from. What a good fortune and a great honour to have worked side by side with the brightest, the bravest, the most open minded few who also appear to be the youngest.

Perhaps the luckiest moment of my career was very early on, when I met Wyn Evans, who became my mentor, my colleague, and eventually my friend

I was incredibly fortunate to receive support from multiple organisations during my academic career. These included George Soros’s International Science Foundation, which funded me during my undergraduate years in the chaotic Moscow of the post-USSR, and the Dulverton Fund, which made study at Oxford possible. There is a special place in my heart for the STFC Fellowship, which came exactly at the moment when I was trying to reinvent myself and do something new. The independence the STFC Fellowship provided was simply invaluable, and it pains me to see the hiatus the Council is going through at the moment. It is thanks to the Royal Society URF that I could enjoy an extended period of calm and focus: to ponder and meditate, to methodically explore dead ends, to get stuck, and eventually to emerge “from the woods” with new ideas.

The ERC grant was instrumental in getting a Streams project off the ground. It helped assemble a critical mass of talent, in the right place at the right time. I am thankful to the Simons Foundation, and to David Spergel personally, for hosting me and my family in New York twice, at the times of both Gaia Data Release 2 and 3. The Simons Foundation’s Center for Computational Astrophysics was a splendid place to carry out the Gaia Sausage analysis and write the discovery paper, and New York City became a home away from home for our family. Returning to Cambridge after the sabbatical, I was able to rebuild my research group thanks to the support from theLeverhulme Foundation.

But perhaps the luckiest moment of my career was very early on, when I met Wyn Evans, who became my mentor, my colleague, and eventually my friend. Immediately I was struck by Wyn’s theatrical and magnetic character, impressed by his sharp and witty mind and, of course, as we sat down to write up our first results, his eloquence. Over decades, Wyn’s office has reliably been my first port of call for a scientific discussion. With his rare gift for conversation, he has time and again helped me strip away the clutter and get to the core of an idea. I hope that one day I can grow to be as wise, as generous and as patient a mentor as Wyn.

I have not taken this journey by myself. My partner Natalia has shared it with me every step of the way. When I am asked where I feel most at home, the answer is not geographical: home is next to her and our two kids, Mark and Claudia.