शैडोइंग अभ्यास: Inside a Hong Kong Hackathon [Full Documentary] - वीडियो के साथ अंग्रेजी बोलना सीखें

अभ्यास मोड चुनें

हर मोड पूरी स्क्रीन का अभ्यास कक्ष खोलता है।

456 वाक्य

यदि वाक्य बहुत छोटे या लंबे हों, तो समायोजित करने के लिए Edit पर क्लिक करें।

How satisfied are you with what you accomplished this weekend?

Zero, five, five.

I love it.

Great job.

Welcome to C -Hack.

We're here in Hong Kong with more than 250 developers

and builders from more than a dozen countries who are going

to be teaming up to build projects over the next 24 hours.

We're going to judge those projects.

We're going to give them some prizes, and we're going to learn a lot about how

people are building cool projects on a tight timetable so

this is an app that teaches people Cantonese pronunciation yeah you can just record your voice

and it can detect if you have

pronounced correctly and it will correct you if you have wrong pronunciation okay

so you are Jenny and you are Coco how do you know each other just met yesterday

and we form a team yes are both of you from Hong Kong awesome

and tell me a little bit more about your app

and maybe give me some more examples of it working I want to actually try it

and see if I have good pronunciation you can just try out you can start recording

Let's see how how accurate you are

oh congratulations you got 100 % all right well now I

want to eat some you made me hungry we can go buy some awesome yeah bring guidance I

and bribe the judges yeah I will ask them to try try it out yeah we also have other section

but still building like there are some visualizing of the tone

and the six tones of Cantonese and some lyrics

but still building yeah I think lyrics are gonna be really popular

because a lot of people who are not Cantonese speakers like listening to Canton pop yeah any song you have listened

I won't name one on the spot

but yeah I listen to a lot of it all right yeah

so we will also recommend some good Canon pop for people who want to learn Cantonese

and at the same time listen to the music okay yeah I think that's a really cool feature

if you can essentially like have them sing along

and then also like grade their tones

and their pronunciation oh sure we can add this function it

is a good idea yes awesome keep up the great work tell me your name

and where you're from my name is Serena I'm from HQ physics I

specialize in machine learning my name is Brian I'm from Vietnam

currently work in hong kong for a tech company here i'm

ben i also work in a tech company here in hong

kong with brian over here i am josh from the philippines i'm currently a economics

and biotechnology student at hkust i'm carmen i'm from hong kong

and i specialize on vibe coding 3d low poly experiences

and working in ai transformation i'm hyped to learn a little

bit more about what you all are building brian can you

just tell me kind of like fundamentally what the problem

that you're solving is

and for whom all right thank you we are here for the hackathon building for a company called Apple Park

so basically it's a customized

recommendation system that will help the customer have a better flow in choosing the dolls

and the toys for their kids I think the problem

that we encountered is for a lot of people who are

buying gifts for someone else's children they want to show

that they know the child and that they are empathizing with the parent in in the mission to develop the child.

So this recommendation engine that we created is more conversational so that we get to see what are the idiosyncrasies that,

you know, the gift giver wants to express in the gift

and give them the recommendations that resonate with the preferences that they have.

We're here with Fozzie, Sherry, Ricky.

They are building Tinder for Loans.

So we are basically lonely, we are just Tinder for Loans.

So how it works is

that we have the plan generated by the AI with our story given

and the amount of money we need so how it works is that when we click submit

it shows the video where we can submit like a yc video all the applicants can submit it

so how it works is that the lenders can go through the borrowers video

and then see whichever video they like and then

if you click it it shows you the information and

if you swipe right it shows accept and then

if you go down and then if you swipe left it's yes rejects

so that's what basically happens awesome so basically they can just watch your video and

if it's any point they decide yeah I don't want to

fund this I'm not gonna give them a loan they swipe it out of the way

and move on to the next one that's basically how it works

and I guess the idea is this helps people who are applying for loans

because it's just like lower friction for them to get things

in front of potential loan makers yeah also helps the lenders to get more details of the borrowers

and see how much they want

and we can also show there are other histories as well

using blockchain have built this the audit logs they can view through

that and they can see the details and

if they can want to lend them

or not we're here with jeffrey joe from baltimore who flew in to help organize

and run mentor mates hackathon see hack how's everything going with you it's good it's hectic

but it's it's really good we have some really cool projects

really looking forward to see what everyone's building what everyone has

completed basically today awesome how many submissions have you had

so far do you know we have over 20 I think we're nearing 30 something submissions

so 30 teams have submitted

which is great that's awesome maybe you can tell me just

a little bit about what the judging process is gonna be like here yeah

so mainly when we judge hackathons depending on the track we have a couple of different things

that we look for so for example startup community

and industry tracks we're looking for whether for example for startup whether your your idea is venture scalable

when we're looking at industry we're looking at whether you're solving a real problem

and you're solving it correctly for community we're looking at sort of the value you're adding to a non -profit

or potential community effort.

So our judges will be checking those criteria.

For the first round, it's going to be a two -minute pitch video.

All the judges will look at that, and then we'll select the top three from each track to move on.

The people who move on will be able to present their

100

project live to all of the audience as well as the judges, and we'll select one winner from each track.

101

Awesome.

102

So there are going to be three winners tonight.

103

Is there like a grand prize or anything like that

104

or is it just like uh each track prize is the

105

same um we have a couple prizes from the sponsors from our lovely sponsors

106

that will be in addition to those right on sounds really cool

107

and i am honored to be one of those judges to

108

be able to like uh determine the fate oh no this

109

is gonna be awesome i'm hyped to watch some of these

110

videos let's go watch some of those i'm gonna run some

111

of those right now for you good people yeah thank you

112

again quincy for judging uh really looking forward to seeing what everyone's built.

113

This is a 24 -inch pizza right here in Hong Kong.

114

You usually don't see pizzas like this outside of New York City, but we're going to dig into one right now.

115

Ruby, why do you do this?

116

Angela, can you tell us a little bit about what you're building and for whom?

117

Among the three tracks, we choose the Apple Park kits,

118

and our pin point is that among all the dolls of our products, the parents don't know which product they should choose for their kids.

119

So what's our solution?

120

It's that we customize stories for every single product so that the parents would know

121

what specific doll is more suitable for the kids.

122

And how are we going to do that? it's that we're going to have LLMs similar to ChatGDP, a chat box where the parents could put in their kids information

123

and we will generate a story and also have the matching dolls suitable for that kid.

124

We're here with Connor from South Africa,

125

Cindy from Hong Kong, and they are building a tool to help recognize volunteering at a charity, right?

126

Yeah, that's That's right.

127

Maybe you can tell us a little bit about the problem

128

and how you're solving it yeah so basically um

129

so what we've been noticing is

130

that like a lot of angel have their own system

131

and basically the reason we came up with this idea

132

so my friend work in our marketing agency and we noticed

133

that a lot of angel pay a huge amount of money for a marketing agency to recruit like volunteering

134

and sponsor to join the event so basically we're building a platform

135

that we can unify everything keep the cost as cheap as possible

136

so that like all the other people can receive the money

137

that they should deserve rather than just spending all it on marketing

138

so that's something that we are building awesome

139

and what has been the most fun part of building this so far here at CHAC?

140

Taking the idea at conception from six hours ago to having a live demo ready, it's been a lot of fun.

141

I mean, neither of us are technical.

142

She comes from marketing, I come from finance.

143

Never touched code before, never even downloaded like Visual Code Studio or anything like that.

144

But with AI, anything's possible.

145

So So it's been a lot of fun to just turn the idea into an MVP, essentially, in six hours.

146

Right on.

147

Keep up the great work.

148

I'll come back and check.

149

You're going to continue to improve and fine tune this in preparation for judging, right?

150

That's the goal.

151

Hi, I'm Quincy, founder of Free Code Camp.

152

Who are you?

153

Hi, my name is Chinat Yu.

154

It's nice to meet with you.

155

I'm an AI hacker, rapper, builder, because that is what I do.

156

Quincy from Free Code Camp, coming to Hong Kong to leave his stamp

157

because with mentor mates we build better hackathons for us all

158

together we're here c hacks with chinat the organizing force behind

159

the hackathon uh founder of mentor mates who has now run

160

three hackathons this is the second one in hong kong the first one in hong kong

161

which i had the pleasure of attending 50 participants this year

162

250 how do you achieve 500 growth with within the space of one year between two hackathons?

163

I think that if you run great events, you would naturally attract people to kind of gather and help you to grow that event even more.

164

What was really interesting was between last year and this year, so many people in and across Hong Kong, beyond Hong Kong, actually reached out to me, being like, hey China, can I run this hackathon with you?

165

I saw the amazing work that you've been doing here with kind of Quincy, as well as some of the folks locally in Hong Kong.

166

What can I do to kind of help and support?

167

And so this year we're really excited, like you know we have a lot of amazing community partners including angel wyelite

168

and many others local

169

and the community who are actually you know putting out the

170

effort to really gather the folks to get this hackathon up

171

and running you are here in hong kong born and raised

172

and you're a huge believer in the hong kong tech ecosystem

173

what are some reasons people who have been sleeping on hong kong should get over here

174

and explore potentially like creating a startup here

175

or potentially investing in startups here

176

so i think right now there's a huge conversation around like

177

where should i build my ai where is the future of ai is it in china

178

or u .s and the answer is is

179

that the future of ai is driven by both china

180

and u .s and there is no better place to discover

181

that here in hong kong because we are such a global city we're

182

so close to you know shenzhen and the greater bay area in mainland china

183

but at the same time we have this global international

184

and are able to kind of access AI models and AI talent and AI infrastructure from abroad as well.

185

And so being this kind of natural bridge of connecting the two places, Hong Kong is a natural place of where to kind of find where East meets West.

186

We're here with Michael, Hong Kong local, who is building a CRM tool,

187

like customer relationship management tool for local small businesses that doesn't assume

188

that they have like a giant marketing department or anything like

189

that talk a little bit about what you're building

190

and how you're building it i believe

191

that our conversations should be should not be the training data of the next ai

192

and we always want our private conversations to become pretty private

193

and never send to the big text

194

and this is why i have built raft this is a

195

local ai focused way as a customer resource management tool

196

that connects directly to your whatsapp or your telegram

197

which is the tools that is generally mostly used by Hong Kong SMB owners

198

and it will automatically review your chats using completely local AI

199

which is like completely processed locally and never sent to cloud

200

and it will be able to for example suggest your replies to the video customers

201

and also have a brief summary on your relationship with them

202

and also extract memories from them so

203

So it's into like a hundred x your interaction with your customers

204

and most importantly it is completely built okay it's completely running

205

locally it can run on any of the like more than the recently manufactured computers

206

and even efficiently on there without consuming too much resources using the Gemma for E4B local models

207

which means that every single thing will be completely stored locally

208

and never process by the cloud and it keeps our conversation safe

209

which we find the privacy thing is increasingly being concerned in the modern era of AI what are you building

210

if I understand correctly it's gonna use like OCR on the bill

211

and then once it understands the bill it can apply tax

212

who is the ultimate end -user of this like who would buy this application and the

213

businesses that have the needs to manually input data

214

and they will like to have a unified workflow for the

215

junior employees to work in a unified UI for validating the buildings and to digitalize them.

216

Would this be like for like a property management office,

217

like a landlord or anybody who has junior employees that are looking at any document or specifically are you focusing on bills.

218

The current flow is targeting bills only because we have a workflow to

219

and it's just like a rubbish beans to let the employees to drop all the bills into this workspace

220

and then he can refill it in an organized way.

221

It's actually minimized the error of manual data inputs

222

and it makes the whole process more efficient

223

and organized all right terry uh tell me a little bit

224

about what you're building what problem you're trying to solve

225

and how you're solving it sure the problem facing the company is

226

that it takes some it takes much time for some individual

227

users to write the loan to write the checklist to submit their application

228

and now we build an ai assistant alone we can to finish everything with simple chats.

229

And the AI system will evaluate all the things that the individual user provided and give feedback to the administrator.

230

And the administrator could check everything from that website and don't need to chat with the users.

231

hmm homemade cookies nearby china's mom

232

how satisfied are you with what you accomplished this weekend

233

we are super satisfied for what we have accomplished

234

and we think our project could you know change the industry

235

yeah it's about insurance awesome I love your confidence keep it

236

up I would say I'm pretty satisfied with it we have completely implemented local AI layer of the raft

237

which means that right now you can have a local AI to generate all of your social replies

238

and get your memories and most importantly it never leaves your commuter it's just as safe your messages.

239

That is, save your WhatsApp messages.

240

And everything you saw on the computer, it never leaks to the cloud.

241

Awesome.

242

Congratulations.

243

It's satisfying, obviously.

244

We made like a project.

245

It was chaotic at first, but like eventually we gradually solved many problems and we are super satisfied about it.

246

We just submitted our project.

247

It's a start -up track.

248

how does it feel to have submitted your project it's very

249

exciting I mean you can see my first hands okay and this is our demo

250

here's over demo like we generate a little video like show how to use our from our app.

251

Very cool.

252

Are you submitting?

253

Are you done?

254

So we had an issue with our submission.

255

Basically, unfortunately, we couldn't get our Vercel online.

256

So it's live on localhost, I just couldn't get it onto the Vercel.

257

So currently we're uploading a video of the localhost onto YouTube to then just add it into our submission, but it might be too late.

258

i will let you focus on that uh good luck say your name and where you're from igor russia

259

helena russia how does it feel to have submitted the project you spent the last 24 hours working on?

260

It's a bit exhausting, to be honest.

261

Yeah, we barely had any sleep.

262

So, yeah, it was a bit tough.

263

What about you?

264

I'm fine.

265

We're deploying live on Vercel, so you guys can access it.

266

Okay, so they are in the process of uploading and deploying and submitting.

267

Sorry, my teammates are busy.

268

Because we're trying to deploy the prototype that we have on Vercel right now, and it's 3 .30 exactly the deadline time, so we are in quite a hurry.

269

She's narrating our stress levels.

270

Yeah, narrating our stress levels at the very end of the hackathon.

271

You don't want to submit your .in file.

272

Literally.

273

Oh, there it says environment variables.

274

Yeah, because you need it to work.

275

Yeah, that's true, that's true.

276

Okay, so you tried to click import in.

277

Let's see.

278

Maybe it didn't even have an environment to do it.

279

It probably won't work without a .in file though.

280

Let's try it.

281

No, let's just see what happens.

282

Congratulations.

283

Oh wait, it's working.

284

It's working.

285

Yeah, okay.

286

Oh, I love it.

287

Oh!

288

Wait, wait, let's try it.

289

Test it out first.

290

Okay, okay, okay.

291

Can we submit it though?

292

Yeah, let's submit it first.

293

Let's submit it first, right?

294

Show the link.

295

Okay.

296

Where, how do you?

297

It's not now.

298

Yeah just make sure it works before you submit I think they'll probably give you like an extra minute

299

or two okay

300

let me go in get out of the incognito mode

301

and paste it into our slide deck.

302

We need to share this presentation and where's the submission?

303

Okay, there.

304

Okay, so now we need to share the link to this.

305

Rename.

306

It's okay.

307

Share with me and then I'll give it to the organ.

308

I'll give it to China.

309

They're in the process of submitting right here.

310

Alright, we can probably leave him alone.

311

We can come back later.

312

We don't know why we can't select the correct track.

313

So we've got to start collecting the application form.

314

Oh no!

315

These two gentlemen over here, they're having trouble selecting their track to submit.

316

Alright, we're watching some real -time QA.

317

How satisfied are you with what you accomplished this weekend?

318

Zero, five, five.

319

I love it.

320

Great job.

321

Very satisfied.

322

I learned how to code.

323

I was late.

324

I still learn and I'm now able to make something.

325

I made a map overlay for all the natural resources and fertile lands.

326

Yeah.

327

That's sweet.

328

And you did this using like a code harness, like cloud code? did this using cloud code

329

and from the token mart

330

that the organizer gave us congratulations thank you well the final project wasn't finished really

331

but I got to learn a lot and there were a lot of experts

332

so I'm very satisfied with this event awesome congratulations keep keep building you were working on the you

333

thing right did you finish it okay well you

334

keep working on it outside

335

of the hackathon I will

336

tell you I have done hackathons where I could not finish the project in time

337

and I didn't have anything to submit it's fine keep working

338

on it yeah how satisfied are you with what you accomplished this weekend incredibly satisfied

339

because we made the impossible possible i think we cooked we

340

managed to clutch our project really well literally the last few seconds before the submission deadline

341

and i'm really proud of the entire team

342

and what we've been able to make

343

so yeah let's just see what's in store for everyone awesome

344

congratulations thank you thank you thank you they got it submitted

345

awesome congratulations and now we're ready to cover the judging process let's go check it out

346

I do think the actual UI obviously was filled with plotisms and stuff, but it was fine.

347

implementation likelihood like i don't think there's anything stopping this from

348

being a real product yeah profit maybe a four all right

349

yeah i mean i don't know like yeah it's an imagined problem

350

that i cannot relate with yeah i don't have any empathy for

351

that i talked to this guy

352

considering the fact that they didn't even know how to use Claude and stuff like that i think

353

i think execution is probably like a four yeah market support

354

too this is uh this is interesting yeah i mean i can see like

355

if you're in like you want to learn how to do like commodities trading

356

or something that might be like a really helpful tool i don't see you coming back

357

and using it multiple times yeah like how much is there

358

really to know like yeah after you've used it a few times you'll probably figure it out yeah

359

so i think the main challenge is scalability i'm not sure

360

about the uh market off like long term the presentation is pretty good this is

361

so quad quad is like really bad

362

and just like having excessive explanatory text right like

363

if you look here it's like text and then it's text

364

and then it's it's more text like on an actual squeeze

365

page yeah there would just be one heading out it wouldn't

366

be like three things yeah it's just like little things like that that drive me crazy

367

We are the first finalist he presenting, everybody give a big round of applause to Webb!

368

Today I want to present you, Red.

369

Your completely private, fully local second grade of photo -social using AI.

370

What it does is, you connect your WhatsApp or Telegram to it, and then it will automatically drive your replies to your friends or customers.

371

So this is the future of social.

372

If you have any U .S account present that has too much people to talk with, this is for you.

373

Thirdly, this is time to open source.

374

And to be fair, this is connected to the WhatsApp Red Protocol, which is actually out of terms of service.

375

However, our commercial plan will be integration with proper reference, using like for the matter business, or like a WhatsApp business.

376

And this kind of integration actually requires a real business, you know, a real solution provider,

377

and we are going to be it.

378

So we are basically playing the Harms Agent Playbook, right?

379

We sell an open source solution and then we provide integration that requires a cloud.

380

We have built our Canlis 101.

381

Can I tell people to learn Canlis?

382

I don't know how many of you know Canlis or not.

383

Can you raise your hand if you do not know Canlis?

384

Oh, many of us, many of you, so you can definitely check it out.

385

Because we have different functions, we can turn it into English, and so you can understand what is going on.

386

In Japanese, there are a 9 -tone, and we can simplify it into 6 -tone, and we can listen to the pitch.

387

Let's play the pitch.

388

Da, da da da, da da da da da da da da da da da

389