Javier — Blog

From the Nucleus 6 to the Nucleus 8

2026-08-05T00:00:00Z

Today I upgraded.

Nine years with the same implant, the Nucleus 6, and today I finally switched it for a Nucleus 8.

I didn't know what to expect.

I thought it would be a small improvement.
Noticeable, maybe, but manageable.
Something that would feel like a newer version of the same thing.

It wasn't.

The change was radical.

And that's what I still haven't fully processed.

For nine years I got used to hearing a certain way.

I learned to live with that.

I learned not to ask for more, because I didn't know there was more to ask for.

You get used to what you have, even when what you have is limited.

And today, within minutes, I understood that I had spent nine years hearing less than I could actually hear.

That's hard to sit with.

It's not just "now I hear better."

It's realizing everything I didn't know I was missing.

Sounds I didn't notice were gone.

Details I didn't even know existed.

I never thought I needed this.

I didn't even have it in my head as a real possibility.

And now that I've tried it, I can't think about going back.

It's strange to feel this way about a device.

But it's not the device.

It's what the device gives back to me.

I still have adjusting to do.

The brain needs time to relearn how to process all of this.

But the first day already made one thing clear:

nine years thinking I was hearing well,

and only today did I understand how much I was missing.

Nine years, a new implant

2026-08-03T00:00:00Z

Before my accident, hearing well didn't mean anything special.

It was just something that happened on its own.

I didn't appreciate it.
I didn't think about it.
It was just there.

After the accident I lost my hearing.
And that's when I started to understand what it really means to hear.

It's been nine years since I got my first implant: a Nucleus 6.

Nine years hearing the world through that device.

Nine years getting used to a different way of listening.

This Wednesday I'm switching it for a Nucleus 8.

And for days I've been feeling something I didn't expect: excitement.

It's not the kind of excitement I can easily explain to someone who always heard well.

For most people, a Wednesday is just a Wednesday.

For me, this Wednesday
is different.

It's a new device.
It's new technology.
It's the possibility of hearing a little better than I've been hearing for almost a decade.

And that has me thinking about something I don't think about often:
everything I lost that night, and everything I gained afterward.

I lost the way I used to hear before.

I lost taking for granted something as simple as a conversation, a song, background noise.

But I also gained something else: the ability to get excited about something as small as a Wednesday.

Before the accident, I never would have gotten excited about a device.
About an upgrade.
About hearing "a little better."

Now I do.

And it's not that I didn't appreciate hearing before.
It's that I didn't know there was something to appreciate.

You don't value what you take for granted.
I didn't value hearing well because I never had to.
It just happened.

Now every improvement, no matter how small, feels huge.

I don't know exactly what's going to change on Wednesday.
I don't know if I'll notice the difference right away, or if it'll be something that settles in slowly, the way it happened with the Nucleus 6.

But I know I'll be there, waiting to hear the world a little differently.

And I know that excitement, the kind you feel waiting for something as simple as hearing better,
is also something the accident left me with.

I didn't ask for it.

I wouldn't have chosen it.

But there it is.

The Dinner Table Syndrome

2026-04-06T00:00:00Z

The “dinner table syndrome” is a real thing. It’s a term people use, and I relate to it a lot.

For me it’s simple: being at a table with a lot of people, everyone talking about different things, and not being able to follow the conversation.

Voices overlap. They stack on top of each other. Everything happens at the same time.

And at some point, you just stop understanding what’s going on.

It’s not about attention. It’s not about being distracted.

You just can’t process all of it.

That’s when it gets frustrating.

Because you’re there, but you’re not really part of the conversation.

A lot of times what I end up doing is leaving. I just get up and go somewhere else.

From the outside, that can look rude. Like I don’t care.

But it’s not that. It’s just that I can’t keep up anymore.

There’s also something that happens before I even sit down.

When I see a table with a lot of people, I choose who to sit next to.

It has to be someone I trust: my brothers, my parents, someone I don’t have to try so hard with.

Because yeah, having a conversation now takes more effort than it used to.

It’s not obvious at first, but you feel it.

That’s why a lot of times I’d rather do something else: check my phone, watch something, just disconnect a bit.

It’s easier than trying to follow something that I already lost.

I don’t see it as something dramatic.

It’s just one of those things that changed.

And now it’s part of how I experience these situations.

Back to Noise

2026-03-22T00:00:00Z

Before my accident, I liked going to clubs or places with loud music.
It did not bother me.
It was just part of the atmosphere.

After the accident, that changed a lot.
Between the hearing loss and the cochlear implant, loud music started to stress me out.
Not only because of the volume,
but because of everything that comes with it:
people talking at the same time,
mixed noise,
the effort of trying to understand anything inside that chaos.

For a long time, going to places like that exhausted me quickly.
I would start feeling uncomfortable,
overloaded,
as if instead of enjoying myself I was only enduring it.

Because of that, I stopped seeing those places as something fun.
They no longer felt natural to me.
They felt difficult.

Today I went to a wedding.
The music was extremely loud,
there were a lot of people,
and in theory it was exactly the kind of place that would have stressed me out before.

But I realized something:
loud music does not bother me the way it used to.

I do not mean that now everything feels easy,
or that hearing in those environments is simple.
But I clearly felt that something had changed.
I did not go into that immediate state of tension.
I did not feel like leaving just to escape the noise.

In fact, I had a good time.

And that surprised me.
Because it was not something I had noticed little by little.
It became obvious today.
As if I suddenly saw an adaptation that had been happening for a while without me fully noticing it.

I guess that is how it works sometimes.
You change slowly,
you adapt slowly,
and you do not notice it until one day you are in the middle of a place that used to weigh heavily on you
and you discover that it no longer feels the same.

I do not know if this means I have completely moved past those kinds of environments.
I do not think it is that simple.
But it did make one thing clear to me:
I have adapted more than I thought.

And for me, that is already a lot.

How I Changed After the Accident

2026-03-21T00:00:00Z

Sometimes I think the accident did not only split my life in two physically.
It also changed the way I exist in the world.

It was not an immediate or clean change.
It was not like I woke up one day as “a different person.”
Over time, I started noticing that many things that used to define me were no longer there,
and that other things had appeared that did not carry as much weight in me before.

Before

Before the accident, I loved going out and partying.
I was always reaching out to my friends to see if anything was happening:
a gathering, a night out, a party, any plan that would keep me from staying at home.

I was very social.
I liked being around people, moving from one place to another, feeling like something was always going on.
I was studying architecture and, honestly, school did not interest me that much.

I did what I had to do, but my attention was somewhere else:
on my friends, on going out, on seeing what plans would come up that weekend or even that same day.

I do not say this with shame or as criticism.
That was simply who I was.
My energy was directed outward.

After

After the accident, that changed a lot.

I stopped being social in the same way.
I no longer felt the urge to look for people all the time.
I no longer had that need to constantly see what plan was out there,
or that same ease moving through gatherings, parties, or large groups.

Part of it had to do with everything that happened.
Part of it had to do with exhaustion.
And part of it, I think, had to do with the fact that I was no longer the same person.

While that was happening, I started applying myself much more in school.
It feels strange to think about now, because it was almost the opposite movement from who I had been before.
What used to matter little to me started mattering a lot.

I also changed careers.
I left architecture and moved into programming.
It was not a minor change.
It was a major decision, and at the time it felt like starting over,
but also like moving closer to something I could truly sustain.

I became extremely focused.
I studied with a discipline I had never had before.
I took my degree seriously.
In the end, I graduated with high academic performance, so I did not even have to present a thesis.

From the outside, that sounds good.
And yes, it was a real achievement.
I am glad I did it.

But at the same time, I also went through a period of depression.

I say this because stories of change are often told in a way that feels too neat:
as if someone goes through something hard, matures, becomes disciplined, and that is the end of it.
In my case, it was not like that.

Yes, I became more disciplined.
Yes, I changed careers.
Yes, I got good results.
But that did not erase the difficult part.
It did not erase the isolation.
It did not erase the sadness.
It did not erase the period when I simply felt bad.

I think both things existed at once:
on one side, a version of me that was much more focused and disciplined;
on the other, a period of depression and distance from other people.

I do not know which version was “more me”

I do not know if the accident turned me into a different person
or if it only revealed parts of me that had been hidden before by noise, partying, and inertia.

Sometimes I miss things about who I was before:
the ease of being around people,
the lightness,
that way of living without thinking so much.

But I also recognize that afterward I developed things I probably would not have developed otherwise:
discipline,
focus,
consistency,
and a much more serious relationship with my work and with what I want to do.

I am not writing this to say that one version was better than the other.
Only to make one thing clear:
there are events that do not only change your life on the outside,
they also reorganize you on the inside.

And then you have to learn how to live with that new mixture.

When you can’t tell how loud your own voice is

2026-03-07T00:00:00Z

There’s something curious that happens when you don’t hear well: you don’t always know how loud you’re speaking.

This used to happen to me a lot before getting my cochlear implant, when I was only using hearing aids.

In quiet places it wasn’t very noticeable. But in crowded restaurants it definitely was.

Once, when I was living in Ecuador, I went to a restaurant with my mom and my dad.

The place was packed. People talking everywhere, plates moving, music in the background, waiters walking around — everything mixed together.

At some point I wanted to order something to drink.

So I called the waiter.
—Hey!

To me it felt completely normal.

A reasonable volume. Polite, even.

But the waiter turned around as if someone had called him from the other side of the restaurant.

My mom looked at me and said:
“Don’t yell at him.”

I was confused.

I didn’t feel like I had yelled. In my head I was speaking perfectly normally.

Something similar happened to me later at a taco place.

When we finished eating I raised my hand to ask for the check.

—The check, please!

The waiter jumped a little.

Not a small flinch. An actual jump.

Again, I didn’t feel like I was yelling.

That’s one of the strange things about not hearing well: you lose a bit of reference for your own voice.

For everyone else, you’re shouting.

For you, you’re just having a perfectly calm conversation.

What people have told me about Play & Learn

2026-02-21T00:00:00Z

Some things don’t show up in metrics.

They’re not in dashboards, analytics panels, or usage reports. But when they happen, you know the project is moving in the right direction.

Recently, I was told something that genuinely made me proud.

A mother wasn’t sure whether she wanted her child to use the app. She downloaded it on her own phone first to test it. She liked it. Then she called to ask if it could be installed on two devices.

That kind of validation is different.

It’s not a social media “like.” It’s an adult testing the tool before giving it to their child. It’s judgment. It’s trust.

The images 👌

I was also told something simple: “the images 👌”.

It sounds small, but it isn’t. The app relies heavily on visual clarity, structure, and appeal. If the images work, the experience flows.

What it really means

Play & Learn isn’t just a collection of mini-games. It’s structured. It’s level-based content. It adapts. It tries to make learning English feel lighter.

Knowing that parents test it first, that they like it, and that they ask how to use it on multiple devices… means the experience communicates trust.

And honestly, that matters more than any metric.

When Addressables worked on Android… but not on iOS

2026-02-07T00:00:00Z

In one of the projects I worked on, we used Unity Addressables to download dynamic content from Azure. The workflow was straightforward: build the application, build Addressables per platform, and download the content at runtime.

For a while, everything worked without issues.

Until someone generated a new build.

The problem

One of my teammates generated:

an application build
an Addressables build

When testing the app, something odd showed up:

On Android, Addressables downloaded and worked correctly.
On iOS, the app launched, but the remote content simply didn’t load.

The issue was shared with the team to see if anyone could identify the cause.

What it wasn’t

Before touching any settings, we checked the usual suspects:

Correct Azure paths
Properly built bundles
No obvious console errors
No network issues

None of that explained the failure.

The strangest part was that the exact same content worked on Android, which almost always points to a platform or build configuration issue, not a logic error.

Looking closely at the iOS build

iOS is usually more aggressive than Android when it comes to code optimization. While reviewing the Player Settings, I noticed a setting that often goes unnoticed:

Code Stripping

On iOS, it was set to High.

The real issue

With Code Stripping set to High, Unity removes classes and methods it considers “unused”.

The problem was subtle:

Some scripts were not directly referenced
They were only used when an Addressable was downloaded and loaded
At build time, Unity marked them as unnecessary
On iOS, those scripts never made it into the final build

The result:

Android kept working
iOS failed silently when loading Addressables

The fix

It was a simple change — and easy to miss:

Change Code Stripping from High to Minimal on iOS.

After that:

The scripts stopped being stripped
Addressables loaded correctly again
The behavior matched across platforms

What this bug taught me

Not every bug lives in the code
iOS and Android don’t optimize the same way
Aggressive code stripping can clash with Addressables
Sometimes the problem isn’t what breaks, but what never made it into the build

These kinds of issues rarely show up in tutorials, but they happen all the time in real projects. That’s why they’re worth documenting.

The game I’ve been developing (intermittently)

2026-01-29T00:00:00Z

This game has existed for a while… just not consistently.

I work on it intermittently, mostly on weekends and, when possible, on weekday nights. It’s not a project with clear deadlines or aggressive sprints. It’s something I slowly push forward, whenever time and energy allow.

Still, I wanted to document something that already feels like real progress.

Early development (very early)

The game is in an early stage. There are no finished levels, no balance, no final enemies. I’m building the basics: how the character moves, aims, shoots, and how all of that feels together.

Weapon switching

One of the things I really wanted to get right was weapon switching. I didn’t want a simple appear/disappear. I wanted the body to follow the movement and for the switch to feel intentional.

Separating the body was key

The character is built with separated arms, head, torso, and legs. This allows me to blend animations, like walking while aiming, without the character feeling stiff.

Weapon feel

I started adding recoil, muzzle flash, and small visual tweaks. It’s not the final version, but the weapon already feels better when shooting.

Weapon switching and shooting — WIP

Going slow still counts

Progress is slow, yes. But it’s happening. Working on it only on weekends or at night makes every step matter more. For now, that’s enough.

Excerpt: "Step-Down"

2026-01-28T00:00:00Z

CHAPTER 16 – Step-Down

One day Dr. Garza was in the room, along with my mom and my dad.

I was lying there, looking at the ceiling.

I wasn't trying to prove anything. I wasn't focused on moving. My body did what it could. Or it did nothing.

And then it happened.

The foot.

My foot moved.

It wasn't big. It wasn't clean. It wasn't a movie-worthy movement. It was barely a minimal gesture, clumsy, almost involuntary.

But this time, that was enough.

My mom saw it. So did my dad.

The doctor leaned over the bed and touched my foot.

"Can you feel this?"

I said yes. Or I meant to say yes.

Then I moved it again.

From the outside it was barely that: a tiny movement. But for them, it wasn't tiny.

The temperature in the room changed.

The doctor stood still for a second. Then he turned to my mom and hugged her.

"I thought he wasn't going to be able to move his foot again."

My mom cried.

Not like before. Not from fear. This time she cried from something else.

It didn't resolve everything. It didn't erase the rest. But it opened a door.

Note: this is an excerpt. More of the chapter will be published later.

Pros and cons of not hearing

2026-01-28T00:00:00Z

Not hearing is not an abstract thing. It has very concrete effects on daily life.

Some are comfortable. Others not so much.

Over time, I’ve learned to identify which things work in my favor and which ones require constant adaptation.

Pros

When I take off my implant, I hear nothing. That means I can decide when I want to hear and when I don’t.

At night, for example, environmental sounds don’t wake me up. I don’t hear cars passing by, distant horns, or unexpected noises.

The neighbor’s dog doesn’t wake me up either.

Sleep becomes deeper. Not better. More continuous.

In very noisy places, like restaurants or bars, if the music is too loud or I simply don’t like it, I can turn the implant off.

I don’t have to endure the volume. I don’t have to adapt to the environment.

In that sense, the most important thing is this: I have a choice.

Cons

Not hearing also has clear limitations.

One of the most obvious ones is spatial sound. If someone talks to me from behind or from the side, I don’t always know where the voice is coming from.

I have to turn, look around, confirm.

Another drawback is the alarm clock. I don’t hear it.

To be able to wake up, I sleep with a vibrating watch. It works, but it’s a permanent adaptation.

There are also experiences that change completely.

Going to the movies is one of them.

I can’t watch movies in Spanish without subtitles. The dialogue isn’t clear enough to follow without visual support.

That limits which movies I watch and how I watch them.

It’s not better or worse. It’s different.

Dysautonomia in daily life (in my case)

2026-01-22T00:00:00Z

This is not a medical explanation of dysautonomia. It’s simply how it shows up in my day-to-day life and what tends to trigger it in my case.

What triggers it for me

One of the most common triggers for me is stress, especially when I’m very focused on something and don’t take breaks. For example, while programming: if I spend too many hours in a row in front of the computer (sometimes three hours without stopping is enough), I can start to feel unwell.

It can also happen in social situations, especially when there are many people talking at the same time, a lot of noise, or when I have to keep a conversation going for a long while.

Alcohol is another trigger. It doesn’t have to be much: sometimes a single beer is enough to make me dizzy. Because of that, if I drink, I usually pair it with water or electrolytes.

Standing for long periods of time also affects me. I remember this since I was a kid: for example, waiting in line for a roller coaster, being on my feet for too long would already trigger the discomfort.

What I do when it hits

When I start to feel it coming on, what helps me the most is lying down and lifting my legs. If I can, I also try to sleep.

I don’t always fall fully asleep. Sometimes I just close my eyes and stay still. Even if I don’t sleep, simply “shutting down” for a while and resting is usually enough for it to pass.

A quick note

Not everyone with dysautonomia has the same triggers, and not everyone benefits from the same things. This is just what works for me.

When my cochlear implant was activated

2026-01-22T00:00:00Z

The day my cochlear implant was activated, I went with my brother and my grandmother. I clearly remember the clinic.

The doctor was sitting in front of the computer, with the equipment on the desk. She made adjustments, tested things, looked at the screen, and adjusted again. It was a very normal appointment, just with a very specific goal: turning it on.

At some point she asked me:
—Can you hear anything?

I stayed quiet for a moment. It wasn’t a “clean” sound, nor something easy to describe. But there was something.

I told her yes.

My brother was recording me with his phone. In the video you can see that exact moment. When I said yes, he started crying.

I didn’t react much. I stayed focused on listening.

The doctor continued with the appointment: more adjustments, more tests, more questions. At some point my grandmother said something (I don’t remember what) and we laughed a little. It was a calm moment.

Days later I realized something I sometimes take for granted: how much your family carries with you, even when you’re focused on something else.

I don’t say this as a lesson or a nice phrase. Just as a fact.

Nothing spectacular happened that day. We went to the doctor, activated the implant, and went back home.

But that moment stayed very clear in my mind: the doctor’s question, the phone recording, and my brother crying next to me.

Excerpt: "The Ravine"

2026-01-21T00:00:00Z

CHAPTER 3 – The Ravine

I woke up without knowing how much time had passed.

My back was burning, as if someone had lit a fire inside me. My head throbbed with a slow, heavy pulse. The air smelled of dampness, wet earth, crushed plants.

I was cold.
Afraid.

I could hear distant voices, above.

—Are you okay?
—Stay there.
—Don’t move.

I tried to move my legs.

Nothing.

I couldn’t clearly feel the ground, but I knew I was lying somewhere deep. The darkness was so thick it seemed to have weight. I couldn’t make out shapes—only black stains on black.

Then I saw the lights.

Above, between the trees, flashlights were moving. Small circles of light advancing slowly, illuminating fragments of the ravine. The voices came with them, rising and falling, bouncing between the branches.

I understood something with uncomfortable clarity: if I stayed there, they weren’t going to reach me.

I had to move.

Not later.
Not when it was possible.
Now.

I tried to sit up. My body didn’t respond.

So I did the only thing I could do.

I started to crawl.

I didn’t know where. I didn’t know if I was going up or down. I just moved. But not calmly—clumsily, desperately. As if my body knew something my mind hadn’t yet understood. As if staying still meant disappearing.

I moved my arms again and again without thinking, pulling the wet earth with everything I had. My hands sank into the mud. My nails scraped stones and roots. The pain in my back flared with every attempt, but I didn’t stop.

I couldn’t.

As I crawled, the flashlights kept moving above. But they weren’t getting closer.

They were moving away.

The lights grew smaller.
The voices weaker.

That made me hurry.

Note: this is an excerpt. I’ll publish other parts of the chapter later.

How I evaluate AI agents: personality, RAG, and semantic similarity

2026-01-21T00:00:00Z

The part that interests me most about working with AI agents is not “making them answer”, but making sure they answer the way they should. That they stay in character, don’t invent things, and keep a consistent style even when the user changes topics or asks ambiguous questions.

That’s why, beyond building the RAG itself, I also design evaluation pipelines: repeatable tests to measure whether the system actually improves or just “feels better”.

1) Personality: does the agent behave as the prompt says?

When I say “personality”, I don’t mean being funny or charismatic. I mean clear rules: tone, boundaries, safety level, patience level, how much it asks before assuming, and how well it stays within its intended goal.

To evaluate this, I use prompt batteries (scenarios) that stress the agent: ambiguous questions, sudden topic changes, contradictions, provocations, etc. What matters is that these tests remain consistent, so different versions can be compared fairly.

That’s where you start finding strange behaviors: a tiny tweak can make the agent “more obedient” but also more rigid, or cause it to lose naturalness.

2) RAG: does it actually help, or does it just add noise?

With RAG, the risk isn’t only hallucination — it’s also distraction. Sometimes the model grabs the wrong chunk and fixates on an irrelevant detail. Or the system retrieves the right text, but in the wrong order.

Instead of assuming “more documents = better”, I treat it as a hypothesis: change one variable, run the test set, and compare results.

3) Semantic similarity: comparing RAG vs non-RAG

One metric I find especially useful is measuring semantic similarity between responses. For example: the same question, two configurations (RAG vs non-RAG), and then comparing how close each answer is to the expected outcome.

It’s not a silver bullet, but it helps reveal patterns: when RAG truly anchors the response, and when it only makes the model sound more confident without being more correct.

4) What I’m ultimately aiming for

In the end, what I want is simple: an agent that keeps its behavior under pressure, and a RAG that adds precision, not just extra text.

The interesting part is that this work never feels finished. You’re constantly choosing between trade-offs: accuracy vs latency, consistency vs flexibility, and “answer quickly” vs “ask the right question”.

And when you finally think you’ve got it right, you switch models… and start all over again.

Excerpt: "The Party"

2026-01-20T00:00:00Z

CHAPTER 1 – The Party

February 12, 2016 started like any other Friday.

Nothing strange.
Nothing special.

Although I was in a particularly good mood.

Deadpool was coming out that day, and since the morning, at university, we talked about nothing else. I was studying architecture, so we spent way too much time together: long classrooms, tables stained with glue, unfinished models, endless deadlines, all-nighters that left us stupid and happy at the same time. Between classes, in the hallways, while waiting for the professor, we replayed trailers, exaggerated scenes, made bad jokes, made fun of everything.

Life felt light.
Predictable.

I was more excited than the rest because I already had my plan: going to the movies with my brother in the afternoon.

And later that night, it was a friend’s birthday. There was a party.

That’s what many Fridays were like: university, laughter, movies, cheap alcohol, bar hopping when possible. Nothing extreme. Nothing out of control. Life moving forward without asking permission.

Nothing that made me think everything could break in a matter of hours.

In the afternoon I went to the movies with my brother. The film left us laughing like when we were kids: bumping shoulders as we left the theater, repeating lines, exaggerating scenes as if we didn’t want to let them go yet. One of those simple moments that feel solid. Stable. Like everything was in its place.

I ran up the stairs when we got home. I was already late.

I went straight to my parents’ room. They were both there.

—Hey, I’m going to a party. It’s a friend’s birthday —I said from the doorway, without fully stepping inside.

My dad looked at me for a second. Not with suspicion. More like he was measuring something he couldn’t quite explain.

—What time will you be back?

—I don’t know… later. I’ll text you.

My mom looked up.

—Have your brother take you and let me know when you arrive.

—Okay.

Nothing else.
Just that.
The way you say things you believe will always repeat themselves.

My brother drove me. The ride was short, normal, one of those you don’t remember because nothing stands out. He dropped me off at the entrance of the complex where my friend lived: a large black gate. To the side, a small booth with a slightly bored security guard. A few houses lined up on a slope, lit by yellow lights that barely cut through the night. From there, you could already hear the music, muffled, vibrating in the air.

The party was all the way in the back: a house with two gardens connected by stone stairs.

—Good evening, I’m going to Andrés Cervantes’ place.

The guard opened the gate without really looking at me.

Inside, everything was noise and movement. Shouting. Hugs. Raised cups. Music I don’t remember what it was. Alex, as always, clinked his can against mine.

—Let’s go, man.

The vibe was light. Laughter. Everyone talking at once.

I started drinking.
One beer.

That’s it.

And then it happened.

A brief dizziness.
Sharp.

As if the floor moved just a centimeter.
Not enough to fall.
Enough to notice.

My stomach tightened a bit.
My head felt strange, hollow.

Not like getting dizzy from drinking.
More like something wasn’t properly in place.

I took out my phone and called my mom.

—Mom… I got dizzy.

Note: this is an excerpt. I’ll publish other parts of the chapter later.

The accident book: why I’m writing it

2026-01-14T00:00:00Z

I’m writing a book about my 2016 accident because it was an event that split my life in two, and over time I realized I remembered it in fragments: isolated scenes, conversations, sensations, gaps. Writing it is a way of putting all that in order.

I’m not writing it to teach a lesson, or to make it “inspirational.” I don’t want to tell it as a tragedy either. I prefer to describe it as it was and as I understand it now, calmly and without exaggeration.

There’s another reason. When something like this happens, the story stays in the family, with friends, in what people think happened. This book is my complete version, with the details I do remember and also with what is still unclear.

This process isn’t linear. Some days I write quickly, and other days I don’t move forward at all. Sometimes I rewrite the same thing several times until it sounds honest and simple. If something feels too dramatic, I cut it.

This site will work as a place to publish parts of the process:

book excerpts (when it makes sense to publish them)
notes about how I’m writing it
things I’ve learned while trying to remember and organize the story

Not everything I write will end up here. Some parts will stay private, and others I’ll handle carefully out of respect for people who appear in the story. If a detail changes or I correct something later, I’ll correct it without making a big deal out of it.

If you’re reading this: you don’t need prior context or to follow things in order. You can read a single piece on its own. The idea is that each entry stands by itself.

Listening to music with a cochlear implant

2026-01-14T00:00:00Z

Before getting my cochlear implant, music was an important part of my life. I liked discovering new music, listening to full albums, and finding things that were different from what I already knew.

With the implant, that relationship changed. I didn’t stop listening to music, but now I mostly listen to music I already knew from before.

The main reason is that new music takes more effort. Listening to something unfamiliar means more work: understanding the voice, separating instruments, recognizing the rhythm. All of that is more tiring than it used to be.

Music I already know works better. Knowing how a song goes, when the voice comes in, or how the rhythm moves makes listening much easier.

That doesn’t mean I enjoy it less. On the contrary: I really like returning to songs that were already part of my life. I just listen to them in a different way.

I no longer put music on just as background. If I play music, it’s because I want to listen and I know I have the energy for it.

I’m not trying to go back to how I listened before. I don’t feel like music has stopped being important. It’s a different relationship, and for now, this is how it works for me.

What I’m working on

2026-01-05T00:00:00Z

Right now I’m working on building AI agents based on large language models, specifically systems that use RAG (retrieval-augmented generation).

Most of my time isn’t spent writing “nice” prompts, but tuning how the system retrieves information and how that context is passed to the model so it responds the way each project expects.

In AI, there’s almost never a final configuration. There’s always something new: a different model, a parameter that shifts behavior, or a limitation you hadn’t noticed before.

With RAG in particular, a big part of the work is deciding how documents are split before being stored in the knowledge base. Chunk size, overlap, and how much context you preserve make a huge difference in the responses.

There are also many decisions around retrieval itself: what kind of search to use, how many results to bring back, how to filter them, and in what order to pass them to the model. There’s no single correct approach — it depends heavily on the type of information and what the user is actually expecting.

Part of the job is testing, measuring, adjusting, and testing again. Sometimes a small change improves results a lot. Other times it makes things worse, without an obvious reason.

What’s interesting is that you rarely feel “done”. You reach an acceptable point and know that tomorrow, with a new model or a different approach, things will need to move again.

I don’t say this as a complaint. It’s simply how this kind of work functions right now.