9.6 C
New York
Tuesday, March 18, 2025
Home Blog Page 3778

Wolt — Android Interview Expertise — | by Dev D | Jul, 2024


Touchdown an interview with Wolt was each exhilarating and nerve-wracking. I’d admired the corporate for its modern strategy to meals supply and its smooth, user-friendly app design. As an Android developer, I used to be enthusiastic about the potential for becoming a member of their crew and contributing to such a cutting-edge product.

Section 1: TA Spherical

The journey started with a Video screening with a recruiter. A recruiter from Wolt contacted me to debate my resume and general expertise. She requested about my background in Android improvement, particular initiatives I’d labored on, and my motivation for wanting to affix Wolt.

This dialog was pleasant and casual, serving as a approach to gauge my enthusiasm and guarantee my expertise aligned with the job necessities. By the top of the decision, I felt an excellent reference to the recruiter and was keen to maneuver ahead.

Section 2: Technical Interview
The second spherical was 90-minute Interview with the Workers Engineer and Engineering Supervisor.
They requested me about how my current expertise, asking questions on why I select particular strategy and structure, and many others.

They requested some behavioral questions specializing in Group Dynamics.
Know Extra…

Section 3: Technical Task

A number of days later, I acquired an electronic mail with a technical project. The duty was to construct a small Android software to construct in Kotlin.
I acquired 10 days to finish the project, which appeared cheap.

I spent the primary couple of days sketching out the app’s structure and planning my strategy. I made a decision to make use of Kotlin, given its trendy options and robust help throughout the Android neighborhood.

Constructing the app was a mixture of pleasure and problem. I targeted on making a clear, maintainable codebase utilizing the MVVM structure. I integrated well-liked libraries like Retrofit for networking, Glide for picture loading, and Room for native database storage. Testing was essential, so I wrote unit assessments for the ViewModel and integration assessments for the Repository.

By the top of the week, I had a practical and polished app. I submitted it with an in depth README, explaining my decisions and offering directions for working the app.

Section 3: Technical Interview

A number of days after submitting the project, I used to be invited to a technical interview by way of video name. This stage concerned a panel of three engineers from Wolt’s Android crew. They began by asking me to stroll by my answer for the project. I defined my structure, design decisions, and the trade-offs I made because of time constraints.

The engineers had been significantly taken with my use of the MVVM sample and the way I dealt with state administration and asynchronous knowledge loading. They requested detailed questions on my code, reminiscent of why I selected particular libraries and the way I ensured the app’s efficiency and stability. I used to be glad I had documented my thought course of completely, because it helped me articulate my choices confidently.

Section 4: Behavioral Interview

The ultimate stage was a behavioral interview with a Hiring supervisor and a member of the HR crew. This interview was much less about my technical expertise and extra about my match throughout the Wolt tradition. They requested about my experiences working in groups, dealing with conflicts, and managing tight deadlines.

One memorable query was a couple of time I failed on a mission and the way I dealt with it. I recounted an occasion the place a significant launch was delayed because of unexpected bugs. I defined how I coordinated with my crew to establish and repair the problems, communicated transparently with stakeholders, and ensured such issues had been mitigated sooner or later by higher testing and code opinions.

The dialog flowed naturally, and I felt they had been genuinely taken with understanding my character and work model. In addition they gave me perception into Wolt’s work tradition, their emphasis on collaboration and steady enchancment, and what they anticipated from a brand new crew member.

Conclusion

Getting ready for an Android Developer Interview is a time-consuming the duty,

Preserve in Thoughts earlier than submitting the Task:

  • Use newer applied sciences
  • Add Unit Take a look at instances though it was not talked about within the Task description.
  • Use Modular strategy
  • Add ReadMe File

Sharpening Kotlin expertise, SOLID Ideas, Design Patterns, and App improvement.

A lot of the candidates acquired rejected within the App Improvement Spherical the place I may also help you to clear this spherical and provides your applicable suggestions by reviewing the code.
Comply with me and e book an appointment for extra assist.

Validating names in databases with the assistance of Melissa’s international identify verification service


Firms which can be gathering information want to make sure that the information is legitimate as a way to really make good use of it. And ensuring they’ve the right names of their database may also help set up a great buyer relationship by supporting a buyer’s sense of id. 

Suppose again to occasions once you’ve signed up for a service and then you definitely get an automatic e mail that claims “expensive x” as an alternative of your identify, or maybe lists your final identify, not your first. It’s straightforward to fill out a type incorrectly and thus have your info incorrectly listed in an organization’s database. 

When conditions like this occur and an organization reaches out utilizing this incorrect info, it may be unhealthy for the model’s fame. Due to this fact, validating database names might be useful. Validating names, nonetheless, isn’t the best course of. Not like e mail validation the place there’s a particular format an deal with has to observe, or deal with verification the place there’s a set variety of legitimate addresses, the chances for various names are seemingly infinite. 

RELATED: Cleaning e mail lists will assist protect your sender fame rating

Simply inside the US there are a variety of various cultures that names draw inspiration from, a number of methods to spell the identical identify, distinctive characters that may present up, like for example in hyphenated final names, and extra. And people potentialities develop much more once you transfer round to completely different international locations with completely different languages. 

To assist firms validate the names of their lists, the information firm Melissa presents a identify verification service the place it maintains a big listing of identified names, which can be utilized to validate the names in your listing, in keeping with the corporate’s information high quality analyst Tim Sidor.

“Names are usually not a definite information base set, the place it’s static, and every nation or group has a definite set of names or guidelines,” he mentioned. “So names are very fluid, and the legitimate names are altering on a regular basis.” 

The identify verification service works globally too, by validating names in different international locations utilizing key phrases or characters which can be related to particular areas. For instance, within the U.S., Roy might be a primary identify, however in France, it’s most likely a surname, mentioned Sidor. 

“We all know that completely different languages have sure key phrases that characterize sure issues,” he defined. “And sure prolonged characters are a form of a touch to say, ‘Oh, it’s this language. So we had higher parse that identify that manner.’ So all these issues, taken in tandem, permit us to parse international names a little bit bit in a different way, relying on the place they arrive from, the place they originate.”

In keeping with Sidor, due to the character of names and there being infinite potentialities, their listing can’t probably embrace each identified identify that may ever come into existence. “Resulting from lack of ordinary naming legal guidelines and practices, in addition to non-public firms’ willingness to just accept infinite identify variations, there’s no error proof manner of stopping new and by no means heard earlier than names from coming into fashionable lexicon,” he mentioned. “Due to this fact, there are at all times going to be new legitimate names that aren’t acknowledged as such – being new or completely distinctive.”

Nevertheless, simply because a reputation isn’t on the listing doesn’t imply will probably be acknowledged as invalid. In these situations, the entry won’t obtain a “identified identify discovered” flag, but in addition received’t essentially obtain an “invalid” flag.

“Invalid” flags usually get thrown when there’s a vulgarity or in any other case suspicious identify. The corporate maintains an inventory of identified vulgarities in a number of languages in order that it could possibly flag these in the event that they present up within the discipline. 

Some privacy-minded people might not need to give their actual identify once they’re filling out an online type, in order that they put in a pretend identify, like Mickey Mouse, Harry Potter, Taylor Swift, and so on. These names even have an inventory and would possibly lead to a reputation getting flagged as invalid, or not less than flagged as one thing to examine. 

“Mickey Mouse might be not legitimate, and that’s straightforward,” he mentioned. “However Taylor Swift, there could possibly be a couple of Taylor Swift or the precise Taylor Swift, so that you need to flag it as being suspicious after which perhaps confirm it with the deal with or take another motion to find out whether or not it’s actual or not.”

And at last, the identify verification service can filter out firm names which have ended up in a reputation discipline. As an example, Melissa is the identify of the corporate, however it’s additionally a extremely popular woman’s identify. In that case, Sidor mentioned, having the phrase “information” in there too would flag it, however there’s additionally many indicators that one thing could be an organization, just like the key phrases “company,” “firm,” and so on, which Melissa additionally maintains one other listing of. 

“Firms, they’re legitimate contacts,” he mentioned. “Numerous occasions they’re meant to be in your database, perhaps in a separate desk, however they’re legitimate contacts. You don’t essentially need to eliminate them. You simply need to use these key phrases to determine them as such.”

To listen to Sidor speak extra about Melissa’s identify verification service, watch Episode 2 of our Microwebinar sequence on Knowledge Verification, and tune in on September 18th to find out about cellphone quantity verification, October 18th to find out about deal with verification, and November 14th to find out about digital id verification. And should you missed Episode 1 on e mail verification, watch it now on the similar hyperlink.

Salesforce AI Analysis Introduce xGen-MM (BLIP-3): A Scalable AI Framework for Advancing Massive Multimodal Fashions with Enhanced Coaching and Efficiency Capabilities


Massive Multimodal Fashions (LMMs) are quickly advancing, pushed by the necessity to develop synthetic intelligence programs able to processing and producing content material throughout a number of modalities, corresponding to textual content and pictures. These fashions are significantly invaluable in duties that require a deep integration of visible and linguistic data, corresponding to picture captioning, visible query answering, and multimodal language understanding. As AI applied sciences evolve, successfully combining these totally different knowledge sorts has turn out to be more and more vital for bettering AI’s efficiency in advanced, real-world eventualities.

Regardless of vital progress in growing LMMs, a number of challenges persist, significantly within the accessibility and scale of sources accessible to the analysis group. The first subject is the restricted entry to large-scale, high-quality datasets and the advanced coaching methodologies required to create strong fashions. Open-source initiatives typically must catch as much as proprietary fashions resulting from these constraints, which hinders the power of researchers to copy, perceive, and construct upon present fashions. This disparity slows innovation and limits the potential purposes of LMMs in varied fields. Addressing these challenges is essential for democratizing entry to superior AI applied sciences and enabling broader participation of their improvement.

Present approaches to constructing LMMs sometimes contain refined architectures that successfully combine imaginative and prescient and language modalities. As an example, cross-attention mechanisms are generally used to hyperlink these two knowledge sorts, as seen in fashions like Flamingo and LLaVA. These strategies rely closely on large-scale pre-training, adopted by fine-tuning particular duties to boost mannequin efficiency. Nonetheless, regardless of their success, these fashions must be improved, significantly concerning knowledge scale, range, and the complexity of their coaching processes. For instance, the BLIP-2 mannequin, though a pioneering effort, wants assist with the dimensions and variety of its coaching knowledge, which hampers its capability to realize aggressive efficiency in comparison with extra trendy LMMs. The intricate Q-Former structure utilized in BLIP-2 provides additional challenges in scaling up coaching processes, making it tough for researchers to work with bigger datasets.

Researchers from  Salesforce AI Analysis and the College of Washington have launched the xGen-MM (BLIP-3) framework as an revolutionary resolution designed to boost the scalability and accessibility of LMMs. The xGen-MM framework builds upon earlier efforts however introduces a number of key enhancements to beat earlier fashions’ limitations. The framework makes use of an ensemble of multimodal interleaved datasets, curated caption datasets, and publicly accessible datasets to create a sturdy coaching surroundings. A big innovation in xGen-MM is the alternative of the Q-Former layers with a extra scalable imaginative and prescient token sampler, particularly a perceiver resampler. This alteration simplifies the coaching course of by unifying the coaching aims right into a single loss operate at every stage, streamlining the mannequin improvement course of and making it extra accessible for large-scale coaching.

The xGen-MM (BLIP-3) framework incorporates a number of superior applied sciences to enhance the effectivity and effectiveness of multimodal coaching. Central to the framework is a pre-trained giant language mannequin (phi3-mini) paired with a imaginative and prescient token sampler. This mixture permits the mannequin to deal with free-form interleaved photos and texts, which is important for duties requiring a deep understanding of multimodal content material. The coaching course of features a dynamic high-resolution picture encoding technique, enabling the mannequin to successfully course of photos at various resolutions. This technique includes patch-wise encoding of photos, preserving their decision whereas lowering the sequence size of imaginative and prescient tokens. This methodology enhances the mannequin’s capability to interpret text-rich photos and considerably reduces computational necessities, making the mannequin extra scalable and environment friendly for large-scale purposes.

The efficiency of the xGen-MM (BLIP-3) fashions has been rigorously evaluated throughout a number of multimodal benchmarks, demonstrating spectacular outcomes. As an example, the instruction-tuned fashions confirmed excellent efficiency in visible query answering (VQA) and optical character recognition (OCR) duties. Particularly, xGen-MM considerably outperformed comparable fashions in duties corresponding to TextVQA and COCO captioning, attaining scores of 66.9 and 90.6 in 8-shot evaluations, respectively. Introducing safety-tuned fashions has additional enhanced the reliability of those LMMs by lowering dangerous behaviors, corresponding to hallucinations whereas sustaining excessive accuracy in advanced multimodal duties. The fashions additionally excelled in duties requiring high-resolution picture processing, showcasing the effectiveness of the dynamic high-resolution encoding technique.

In conclusion, the xGen-MM (BLIP-3) framework gives a sturdy resolution for growing high-performance LMMs by addressing vital challenges associated to knowledge accessibility and coaching scalability. Utilizing an ensemble of curated datasets and revolutionary coaching methodologies has enabled the xGen-MM fashions to set new benchmarks in multimodal efficiency. The framework’s capability to combine advanced visible and textual knowledge effectively and precisely makes it a invaluable software for researchers and practitioners.


Try the Paper and Undertaking Web page. All credit score for this analysis goes to the researchers of this challenge. Additionally, don’t neglect to comply with us on Twitter and be a part of our Telegram Channel and LinkedIn Group. If you happen to like our work, you’ll love our publication..

Don’t Overlook to affix our 48k+ ML SubReddit

Discover Upcoming AI Webinars right here



Asif Razzaq is the CEO of Marktechpost Media Inc.. As a visionary entrepreneur and engineer, Asif is dedicated to harnessing the potential of Synthetic Intelligence for social good. His most up-to-date endeavor is the launch of an Synthetic Intelligence Media Platform, Marktechpost, which stands out for its in-depth protection of machine studying and deep studying information that’s each technically sound and simply comprehensible by a large viewers. The platform boasts of over 2 million month-to-month views, illustrating its reputation amongst audiences.



#RoboCup2024 – day by day digest: 20 July

0


The Normal Platform Soccer League in motion.

That is the second of our day by day digests from RoboCup2024 in Eindhoven, The Netherlands. Should you missed the primary digest, which provides some background to RoboCup, you’ll find it right here.

Competitions continued throughout all of the leagues right now, with individuals vying for a spot in Sunday’s finals.

The RoboCup@Work league focusses on robots in work-related situations, using concepts and ideas from different RoboCup competitions to deal with open analysis challenges in industrial and repair robotics.

I arrived on the enviornment in time to catch the superior navigation take a look at. Robots need to autonomously navigate, choosing up and inserting objects at totally different work stations. On this superior take a look at, warning tape is added to the sector flooring, which the robots ought to keep away from travelling over. There’s additionally a fancy inserting aspect the place groups need to put an object that they’ve collected right into a slot – get the orientation or placement of the thing barely incorrect and the it gained’t fall into the slot.

The RoboCup@Work enviornment simply earlier than competitors begin.

Eight groups are participating within the league this 12 months. Government Committee member Asad Norouzi stated that there are plans to introduce a sub-league which would supply an entry level for brand spanking new groups or juniors to get into the league correct.

I caught up with Harrison Burns, Mitchell Torok and Jasper Arnold from Staff MiRobot. They’re based mostly on the College of New South Wales and are attending RoboCup for the primary time.

Staff MiRobot from UNSW.

The group truly solely began six months in the past, so ultimate preparations have been a bit traumatic. Nevertheless, the expertise has been nice enjoyable, and the competitors has gone nicely to this point. Like most groups, they’ve needed to make many refinements because the competitors has progressed, resulting in some late nights.

One notable characteristic of the group’s robotic is the bespoke, in-house-designed greedy mechanism on the top of the arm. The group observe that “it has good versatile jaws, so when it grabs spherical objects it truly pulls the thing immediately into it. As a result of it makes use of a linear movement, in comparison with a whole lot of different rotating jaws, it has loads higher reliability for selecting up objects”.

Right here is a few footage from the duty, that includes Staff bi-t-bots and Staff Singapore.

Within the Center Dimension Soccer league (MSL), groups of 5 absolutely autonomous robots play with a daily dimension FIFA ball. Groups are free to design their very own {hardware} however all sensors need to be on-board and there’s a most dimension and weight restrict of 40kg for the robots. The analysis focus is on mechatronics design, management and multi-agent cooperation at plan and notion ranges. 9 groups are competing this 12 months.

I spoke to António Ribeiro, who’s a member of the technical committee and a part of Staff LAR@MSL from the College of Minho, Portugal. The group began in 1998, however António and most of his colleagues on the present group have solely been concerned within the MSL since September 2022. The robots have developed because the competitors has progressed, and additional enhancements are in progress. Refinements to this point have included communication, the detection system, and the management system. They’re happy with the enhancements from the earlier RoboCup. “Final 12 months we had a whole lot of {hardware} points, however this 12 months the {hardware} appears fairly steady. We additionally modified our coding structure and it’s now a lot simpler and sooner for us to develop code as a result of we are able to all work on the code on the similar time on totally different modules”.

António cited versatility and cost-effective options as strengths of the group. “Our robotic is definitely very low cost in comparison with different groups. We use a whole lot of previous chassis, and our options at all times go to the bottom value potential. Some groups have a number of thousand greenback robots, however, for instance, our imaginative and prescient system is round $70-80. It really works fairly nicely – we have to enhance the way in which we deal with it, but it surely appears steady”.

Staff LAR@MSL

The RoboCup@Residence league goals to develop service and assistive robotic expertise with excessive relevance for future private home purposes. A set of benchmark assessments is used to guage the robots’ skills and efficiency in a sensible non-standardized house atmosphere setting. These assessments embrace serving to to arrange breakfast, clearing the desk, and storing groceries.

I arrived in time to observe the “stickler for the principles” problem, the place robots need to navigate totally different rooms and guarantee that the individuals inside (“company” at a celebration) are sticking to 4 guidelines: 1) there’s one forbidden room – if a visitor is in there the robotic should alert them and ask them to comply with it into one other room), 2) everybody should have a drink of their hand – if not, the robotic directs them to a shelf with drinks, 3) no sneakers to be worn in the home, 4) there needs to be no garbage left on the ground.

After watching an try from the LAR@Residence robotic, Tiago from the group instructed me a bit in regards to the robotic. “The aim is to develop a robotic able to multi general-purpose duties in house and healthcare environments.” Except for the robotic arm, all the {hardware} was constructed by the group. The robotic has two RGBD cameras, two LIDARs, a tray (the place the robotic can retailer gadgets that it wants to hold), and two emergency cease buttons that deactivate all transferring components. 4 omnidirectional wheels permit the robotic to maneuver in any path at any time. The wheels have unbiased suspension programs which ensures that they’ll all be on the bottom always, even when there are bumps and cables on the venue flooring. There’s a pill that acts as a visible interface, and a microphone and audio system to allow communication between people and the robotic, which is all completed by way of talking and listening.

Tiago instructed me that the group have talked to loads healthcare practitioners to search out out the principle issues confronted by aged individuals, and this impressed one in every of their robotic options. “They stated that the 2 most important harm sources are from when individuals are attempting to sit down down or get up, and when they’re attempting to select one thing up from the ground. We developed a torso that may choose objects from the ground one metre away from the robotic”.

The LAR@Residence group.


You possibly can sustain with the newest information direct from RoboCup right here.

Click on right here to see all of our content material pertaining to RoboCup.




AIhub
is a non-profit devoted to connecting the AI group to the general public by offering free, high-quality data in AI.

AIhub
is a non-profit devoted to connecting the AI group to the general public by offering free, high-quality data in AI.


Lucy Smith
is Managing Editor for AIhub.



Expertise Route Might Change Over Time


There are occasions when there’s a longtime expertise that undergoes a modification.  There are additionally occasions when totally different expertise approaches compete.  Within the early days of distributed electrical energy, Direct Present (DC) got here to forefront.  It was understood and constructed on data from working with varied sorts of sources, a few of which we variations of the battery.  Alternating Present (AC) had a extra complicated insertion into on daily basis life because of the reality, that AC didn’t have the arithmetic that might allow the creation of a provide system for AC energy.  Nikola Tesla created the concepts and the rationale on how one can develop the AC system in existence.  Reference 1 supplies a background on Nikola Tesla.

The primary car, which was gasoline powered, was patented by Carl Benz [Ref. 2].   The primary electrical energy automobile ran utilizing a lead-acid battery and was launched in 1888 [Ref. 3].  The primary steam powered car was launched in 1769.  The automobiles have been restricted to individuals with important funds, till Henry Ford created the mass meeting line.   By this time, the flexibility to have gas sources accessible and the reliability of the automobile energy system created using the gasoline engine. 

Now we have two related conditions that contain sources of power.  The present development in demand for electrical energy to drive each cloud storage farms and supply sufficient power for the genAI computational efforts are straining the present power system.  The opposite problem is the electrical automobiles and their want for power however require a considerable time to refill the depleted batteries.

The sources of power are talked about first, as a result of with out that supply, batteries cannot be replenished.  If one considers the assorted sorts of environmentally pleasant operation, there are 4 sorts.  Photo voltaic makes use of the power from the solar and creates DC, which will be transformed to AC.  Wind energy employs capturing power from wind transferring a turbine to generate electrical energy.  Hydro energy (dams) use the pure circulate of water in a river or different stream of water, to show generators to create electrical energy. 

Of those 4 sorts, there are solely two which can be steady, water and nuclear.  Granted that water era does require the stream to be flowing.  There was the completion of a brand new nuclear energy plant in Georgia [Ref. 5].  As has been the historical past of nuclear energy, the fee overrun and time delay from the unique estimate.  This energy plant is able to repeatedly supplying sufficient energy for 500,000 properties for 60 to 80 years.  One subject is that it takes too lengthy to acquire all of the permits and really construct a plant. 

A unique sort of nuclear reactor is being developed and are known as Small Modular Reactors (SMRs) [Ref. 6].  The benefits the SMRs are many within the reality the unit is smaller, has the flexibility be to positioned in smaller areas, will be linked with adjoining reactors to extend energy, and might have a lot of the development achieved individually.  At present, there are greater than 80 of those sort items being developed in nations world wide.  There are additionally different sorts of nuclear reactors being developed from analysis achieved within the Eighties.

That also leaves the difficulty of power storage as soon as it’s obtained.  A lot work has been achieved on Lithium based mostly batteries.  Numerous chemical compositions are being investigated to cut back the present subject with Lithium based mostly batteries.  One massive drawback is the time to recharge the batteries.  There may be an alternate storage materials being investigated: iron-air batteries [Ref. 7].  The benefit of the iron-air battery is that it could have a sluggish discharge of days.  That gives the potential for using the sort of battery to seize photo voltaic and wind generated energy and preserve it saved for days, which Lithium based mostly batteries cannot.  Iron is heavy and appears to point that probably the most favorable utility could be stationary energy storage.  Weight apart, the iron-air battery will be recharged with a lot increased energy ranges than different sorts of batteries.  Is that this one other evolution much like DC to AC?    

   References:

  1. https://en.wikipedia.org/wiki/Nikola_Tesla
  2. https://group.mercedes-benz.com/firm/custom/company-history/1885-1886.html
  3. https://www.edmunds.com/electric-car/articles/first-electric-car.html
  4. https://engines.egr.uh.edu/episode/1596
  5. https://www.georgiapower.com/firm/plant-vogtle.html
  6. https://www.iaea.org/newscenter/information/what-are-small-modular-reactors-smrs
  7. https://lnkd.in/e_w_cvh4?trk=public_post-text Llewellyn King Column
Expertise Route Might Change Over Time

About Walt

I’ve been concerned in varied elements of nanotechnology for the reason that late Seventies. My curiosity in selling nano-safety started in 2006 and produced a white paper in 2007 explaining the 4 pillars of nano-safety. I’m a expertise futurist and is at the moment targeted on nanoelectronics, single digit nanomaterials, and 3D printing on the nanoscale. My expertise consists of three startups, two of which I based, 13 years at SEMATECH, the place I used to be a Senior Fellow of the technical workers once I left, and 12 years at Normal Electrical with 9 of them on company workers. I’ve a Ph.D. from the College of Texas at Austin, an MBA from James Madison College, and a B.S. in Physics from the Illinois Institute of Expertise.