Kuchokera ku physics kupita ku Data Science (Kuchokera ku injini za sayansi kupita ku ofesi plankton). Gawo lachitatu

Chithunzi ndi Arthur Kuzin (n01z3), ikufotokoza mwachidule zomwe zili patsamba labulogu. Zotsatira zake, nkhani yotsatirayi iyenera kuwonedwa ngati nkhani ya Lachisanu osati yothandiza kwambiri komanso yaukadaulo. Kuonjezera apo, ndizofunika kudziwa kuti malembawo ali ndi mawu ambiri a Chingerezi. Sindikudziwa kumasulira ena molondola, ndipo sindikufuna kumasulira ena mwa iwo.

Gawo loyamba.
Gawo lachiwiri.

Momwe kusintha kuchokera ku malo ophunzirira kupita ku malo opangira mafakitale kunachitika kumawululidwa m'magawo awiri oyambirira. M'nkhani ino, zokambirana zidzakhala za zomwe zinachitika.

Inali January 2017. Panthawiyo, ndinali ndi zaka zopitirira pang'ono za ntchito ndipo ndinkagwira ntchito ku San Francisco mu kampani Zoonadi ngati Sr. Katswiri wa Data.

TrueAccord ndi njira yopezera ngongole. Mwachidule - bungwe losonkhanitsa. Osonkhanitsa nthawi zambiri amayitana kwambiri. Tidatumiza maimelo ambiri, koma tidayimba mafoni ochepa. Imelo iliyonse idatsogolera patsamba la kampaniyo, pomwe wobwereketsa adapatsidwa kuchotsera pangongoleyo, ndipo amaloledwa kulipira pang'onopang'ono. Njira iyi idapangitsa kuti pakhale kusonkhanitsa bwino, kuloledwa kuchulukitsidwa komanso kuchepa kwa milandu.

Kampaniyo inali yabwinobwino. Mankhwalawa ndi omveka. Utsogoleri ndi wanzeru. Malowa ndi abwino.

Pa avareji, anthu a m’chigwachi amagwira ntchito pamalo amodzi kwa pafupifupi chaka chimodzi ndi theka. Ndiye kuti, kampani iliyonse yomwe mumagwira ntchito ndi gawo laling'ono chabe. Pa sitepe iyi mudzakweza ndalama, kupeza chidziwitso chatsopano, luso, malumikizidwe ndi mizere mukuyambanso kwanu. Pambuyo pa izi pali kusintha kwa sitepe yotsatira.

Ku TrueAccord palokha, ndidatenga nawo gawo pakuyika machitidwe opangira maimelo pamakalata, komanso kuyika mafoni patsogolo. Zotsatira zake ndizomveka ndipo zidayezedwa bwino ndi madola kudzera pakuyesa kwa A/B. Popeza kunalibe makina ophunzirira ndisanafike, chiyambukiro cha ntchito yanga sichinali choipa. Apanso, ndikosavuta kukonza china kuposa china chomwe chakonzedwa kale kwambiri.

Pambuyo pa miyezi isanu ndi umodzi yogwira ntchito pamakinawa, adakweza malipiro anga kuchokera ku $ 150k kufika pa $ 163k. Kumudzi Open Data Science (ODS) pali meme pafupifupi $163k. Imamera ndi miyendo kuchokera apa.

Zonsezi zinali zodabwitsa, koma sizinatsogolere kulikonse, kapena zinatsogolera, koma osati pamenepo.

Ndimalemekeza kwambiri TrueAccord, kampani komanso anyamata omwe ndimagwira nawo ntchito kumeneko. Ndinaphunzira zambiri kuchokera kwa iwo, koma sindinkafuna kugwira ntchito kwa nthawi yaitali pamakina olimbikitsa pa bungwe losonkhanitsa. Kuchokera pa sitepe iyi munayenera kulowera kwinakwake. Ngati si patsogolo ndi mmwamba, ndiye m'mbali.

Kodi sindimakonda chiyani?

1. Kuchokera kumalingaliro ophunzirira makina, zovuta sizinandisangalatse. Ndinkafuna china chake chapamwamba, chachinyamata, ndicho, Kuphunzira Mozama, Masomphenya a Pakompyuta, china chake pafupi ndi sayansi kapena alchemy.
2. Oyambitsa, ndipo ngakhale bungwe lotolera ndalama, limakhala ndi vuto lolemba anthu odziwa bwino ntchito. Monga poyambira, sizingalipira zambiri. Koma monga bungwe lotolera zinthu, limataya udindo. Tikunena, ngati mtsikana pa chibwenzi akufunsa kumene ntchito? Yankho lanu: "Pa Google" likumveka bwino kuposa "gulu lotolera zinthu." Ndinakhumudwa pang'ono ndi mfundo yakuti kwa anzanga omwe amagwira ntchito ku Google ndi Facebook, mosiyana ndi ine, dzina la kampani yawo linatsegula zitseko monga: mukhoza kuitanidwa kumsonkhano kapena kukumana ngati wokamba nkhani, kapena anthu osangalatsa amalemba pa LinkedIn. ndi mwayi wokumana ndikucheza pa tapu ya tiyi. Ndimakonda kwambiri kucheza ndi anthu omwe sindikuwadziwa pamasom'pamaso. Chifukwa chake ngati mumakhala ku San Francisco, musazengereze kulemba - tiyeni tikamwe khofi ndikulankhula.
3. Kuwonjezera pa ine, Data Scientists atatu ankagwira ntchito mu kampani. Ndinkagwira ntchito yophunzira pamakina, ndipo anali kugwira ntchito zina za Science Science, zomwe ndizofala pakuyambitsa kulikonse kuyambira pano mpaka mawa. Chotsatira chake, sanamvetse kwenikweni kuphunzira kwa makina. Koma kuti ndikule, ndiyenera kulankhulana ndi munthu wina, kukambirana nkhani ndi zomwe zachitika posachedwa, ndikupempha uphungu, pamapeto pake.

Kodi chinalipo chiyani?

1. Maphunziro: physics, osati sayansi yamakompyuta.
2. Chilankhulo chokhacho chomwe ndimachidziwa chinali Python. Panali kumverera kuti ndikufunika kusintha ku C ++, koma sindinathe kuzipeza.
3. Chaka ndi theka ntchito makampani. Komanso, kuntchito sindinaphunzire kaya Deep Learning kapena Computer Vision.
4. Palibe nkhani imodzi yokhudza Kuphunzira Kwambiri / Masomphenya a Pakompyuta pakuyambiranso.
5. Panali kupambana kwa Kaggle Master.

Munkafuna chiyani?

1. Malo omwe padzakhala kofunikira kuphunzitsa maukonde ambiri, komanso pafupi ndi masomphenya apakompyuta.
2. Ndibwino ngati ndi kampani yayikulu ngati Google, Tesla, Facebook, Uber, LinkedIn, ndi zina. Ngakhale pang'onopang'ono, zoyambira zimatha.
3. Sindiyenera kukhala katswiri wamkulu wophunzirira makina pagulu. Panali kufunikira kwakukulu kwa ma comrades akuluakulu, alangizi ndi mitundu yonse ya kulankhulana, zomwe zimayenera kufulumizitsa maphunziro.
4. Nditawerenga zolemba zamabulogu za momwe omaliza maphunziro opanda chidziwitso cha mafakitale ali ndi chipukuta misozi cha $300-500k pachaka, ndidafuna kupita munjira yomweyo. Sikuti izi zimandivutitsa kwambiri, koma popeza amanena kuti izi ndizochitika wamba, koma ndili ndi zochepa, ndiye kuti ichi ndi chizindikiro.

Ntchitoyi inkawoneka yotheka, ngakhale osati mwanjira yoti mutha kulumphira ku kampani iliyonse, koma kuti ngati mufa ndi njala, zonse ziyenda bwino. Ndiko kuti, makumi kapena mazana akuyesera, ndi zowawa za kulephera kulikonse ndi kukanidwa kulikonse, ziyenera kugwiritsidwa ntchito kukulitsa chidwi, kukumbukira kukumbukira ndi kutambasula tsiku mpaka maola 36.

Ine tweaked pitilizani wanga, anayamba kutumiza kunja, ndi kupita zoyankhulana. Ndidadutsa ambiri aiwo panthawi yolumikizana ndi HR. Anthu ambiri ankafuna C ++, koma sindinadziwe, ndipo ndinali ndi malingaliro amphamvu kuti sindingakhale ndi chidwi kwambiri ndi maudindo omwe amafunikira C ++.

Ndizofunikira kudziwa kuti panthawi yomweyi panali kusintha kwa gawo mumtundu wa mpikisano pa Kaggle. Chaka cha 2017 chisanafike panali zambiri za tabular komanso deta yachithunzi kawirikawiri, koma kuyambira mu 2017 panali ntchito zambiri za masomphenya apakompyuta.

Moyo unayenda motere:

1. Gwirani ntchito masana.
2. Pamene tech screen / onsite mumatenga nthawi.
3. Madzulo ndi kumapeto kwa sabata Kaggle + zolemba / mabuku / zolemba zamabulogu

Kumapeto kwa 2016 kunadziwika kuti ndinalowa m'gulu la anthu Open Data Science (ODS), zomwe zinapeputsa zinthu zambiri. Pali anyamata ambiri m'deralo omwe ali ndi zochitika zamakampani olemera, zomwe zinatilola kufunsa mafunso ambiri opusa ndikupeza mayankho anzeru. Palinso akatswiri ambiri amphamvu kwambiri ophunzirira makina a mikwingwirima yonse, yomwe, mosayembekezereka, inandilola, kupyolera mu ODS, kutseka nkhaniyi ndi kulankhulana mozama nthawi zonse za Data Science. Mpaka pano, ponena za ML, ODS imandipatsa nthawi zambiri kuposa zomwe ndimapeza kuntchito.

Monga mwachizolowezi, ODS ili ndi akatswiri okwanira pamipikisano pa Kaggle ndi masamba ena. Kuthetsa mavuto mu gulu kumakhala kosangalatsa komanso kopindulitsa, kotero ndi nthabwala, kutukwana, ma memes ndi zosangalatsa zina za nerdy, tinayamba kuthetsa mavuto mmodzimmodzi.

Mu March 2017 - mu gulu ndi Serega Mushinsky - malo achitatu Dstl Satellite Imagery Feature Detection. Mendulo yagolide pa Kaggle + $20k kwa awiri. Pa ntchitoyi, gwiritsani ntchito zithunzi za satellite + magawo a binary kudzera pa UNet zidasinthidwa. Cholemba pabulogu pa Habre pamutuwu.

Mwezi womwewo wa Marichi, ndidapita kukafunsidwa ku NVidia ndi gulu la Self Driving. Ndinavutika kwambiri ndi mafunso okhudza Object Detection. Panalibe chidziwitso chokwanira.

Mwamwayi, nthawi yomweyo, mpikisano wa Object Detection pazithunzi zamlengalenga kuchokera ku DSTL yomweyo unayamba. Mulungu mwiniyo analamula kuti athetse vutoli ndi kukonzanso. Mwezi wamadzulo ndi Loweruka ndi Lamlungu. Ndinatenga chidziwitso ndikumaliza kachiwiri. Mpikisanowu unali ndi kagawo kakang'ono kosangalatsa m'malamulo, zomwe zidandipangitsa kuti ndiwonetsedwe ku Russia pamayendedwe aboma osati aboma. Ndinakwera kunyumba Lenta.ru, ndi mugulu la zosindikiza ndi zofalitsa za pa intaneti. Mail Ru Group inalandira PR yabwino pang'ono pa ndalama zanga ndi ndalama zake, ndipo sayansi yofunikira ku Russia idapindula ndi mapaundi 12000. Monga mwachizolowezi, zinalembedwa pamutuwu positi ya blog pa hubr. Pitani kumeneko kuti mudziwe zambiri.

Nthawi yomweyo, wolembera anthu a Tesla adandilumikizana nane ndikudzipereka kuti alankhule za udindo wa Computer Vision. Ndinavomera. Ndidadutsa popita kunyumba, zida ziwiri zaukadaulo, zoyankhulana zapamalo, ndipo ndidakhala ndi zokambirana zabwino kwambiri ndi Andrei Karpathy, yemwe anali atangolembedwa ganyu ku Tesla ngati Director wa AI. Gawo lotsatira ndikufufuza zakumbuyo. Pambuyo pake, Elon Musk anayenera kuvomereza yekha pempho langa. Tesla ali ndi Pangano Loletsa Kuwulutsa (NDA).
Sindinadutse cheke chakumbuyo. Wolemba ntchitoyo adati ndimacheza kwambiri pa intaneti, ndikuphwanya NDA. Malo okhawo omwe ndinanena chilichonse chokhudza kuyankhulana ku Tesla kunali ODS, kotero lingaliro lomwe lilipo ndikuti wina adatenga chithunzi ndikulembera HR ku Tesla, ndipo adandichotsa pampikisano chifukwa chazovuta. Zinali zamanyazi pamenepo. Tsopano ndine wokondwa kuti sizinathandize. Udindo wanga pano ndi wabwino kwambiri, ngakhale zingakhale zosangalatsa kwambiri kugwira ntchito ndi Andrey.

Zitangochitika izi, ndidalowa mumpikisano wazithunzi za satana pa Kaggle kuchokera Planet Labs - Kumvetsetsa Amazon kuchokera ku Space. Vutoli linali losavuta komanso lotopetsa kwambiri; palibe amene ankafuna kulithetsa, koma aliyense ankafuna mendulo yaulere yagolide kapena mphotho. Choncho, ndi gulu la Kaggle Masters la anthu 7, tinagwirizana kuti tidzaponya chitsulo. Tidaphunzitsa ma netiweki 480 munjira ya 'fit_predict' ndikupanga gulu la nsanjika zitatu mwa iwo. Tinamaliza lachisanu ndi chiwiri. Tsamba labulogu lofotokoza yankho kuchokera kwa Arthur Kuzin. Mwa njira, Jeremy Howard, yemwe amadziwika kuti ndi Mlengi Fast.AI kumaliza 23.

Pambuyo pa mpikisano, kupyolera mwa mnzanga yemwe ankagwira ntchito ku AdRoll, ndinakonza msonkhano pa malo awo. Oimira Planet Labs adalankhula pamenepo za momwe gulu la mpikisano komanso kuyika chizindikiro kumawonekera kumbali yawo. Wendy Kwan, yemwe amagwira ntchito ku Kaggle ndikuyang'anira mpikisano, adalankhula za momwe adawonera. Ndinafotokozera yankho lathu, zidule, njira ndi zambiri zaukadaulo. Awiri mwa atatu mwa omvera adathetsa vutoli, kotero mafunso adafunsidwa mpaka ndipo kawirikawiri zonse zinali zabwino. Jeremy Howard nayenso analipo. Zinapezeka kuti anamaliza pa malo 23 chifukwa sankadziwa kuunjika chitsanzo ndi kuti sankadziwa za njira yomanga ensembles konse.

Misonkhano m'chigwa pakuphunzira makina ndi yosiyana kwambiri ndi misonkhano ku Moscow. Monga lamulo, kukumana m'chigwa ndi pansi. Koma zathu zinakhala zabwino. Tsoka ilo, mnzake yemwe amayenera kukanikiza batani ndikulemba chilichonse sanakanize batani :)

Pambuyo pake, ndidaitanidwa kuti ndikalankhule ndi Deep Learning Engineer pa Planet Labs yomweyo, ndipo nthawi yomweyo. Sindinachipereke. Mawu akukana ndikuti palibe chidziwitso chokwanira mu Kuphunzira Mwakuya.

Ndidapanga mpikisano uliwonse ngati projekiti LinkedIn. Pavuto la DSTL tidalemba kusindikiza ndikuyika pa arxiv. Osati nkhani, koma mkate. Ndikupangiranso wina aliyense kuti awonjezere mbiri yawo ya LinkedIn kudzera mumipikisano, zolemba, luso, ndi zina zotero. Pali kulumikizana kwabwino pakati pa mawu angati omwe muli nawo mu mbiri yanu ya LinkedIn komanso momwe anthu amakutumizirani mauthenga.

Ngati m'nyengo yozizira ndi masika ndinali waluso kwambiri, ndiye kuti pofika August ndinali ndi chidziwitso komanso kudzidalira.

Kumapeto kwa Julayi, munthu wina yemwe amagwira ntchito ngati manejala wa Data Science ku Lyft adandilumikizana nane pa LinkedIn ndipo adandipempha kuti tidye khofi ndikukambirana za moyo, za Lyft, za TrueAccord. Tinakambirana. Anapereka kuyankhulana ndi gulu lake pa udindo wa Data Scientist. Ndinanena kuti njirayi ikugwira ntchito, pokhapokha ngati ndi Computer Vision / Deep Learning kuyambira m'mawa mpaka madzulo. Anatsimikizira kuti palibe zotsutsa kumbali yake.

Ndidatumizanso kuyambiranso kwanga ndipo adayiyika patsamba lamkati la Lyft. Pambuyo pake, wolemba ntchitoyo anandiitana kuti nditsegule CV yanga ndikupeza zambiri za ine. Kuchokera m'mawu oyamba, zinali zoonekeratu kuti kwa iye ichi chinali chikhalidwe, popeza zinali zoonekeratu kwa iye kuchokera kuyambiranso kwake kuti "Ine sindine chuma cha Lyft." Ndikuganiza kuti pambuyo pake pitilizani wanga adalowa mu bini ya zinyalala.

Nthawi yonseyi, ndikufunsidwa, ndidakambirana zolephera zanga ndi zolephera zanga mu ODS ndipo anyamatawo adandipatsa mayankho ndikundithandiza m'njira zonse ndi upangiri, ngakhale, monga mwachizolowezi, panalinso kupondaponda kwaubwenzi kumeneko.

Mmodzi mwa mamembala a ODS adandipereka kuti andilumikizitse ndi mnzake, yemwe ndi Director of Engineering ku Lyft. Zosavuta kuzinena koma zovuta kuchita. Ndimabwera ku Lyft kudzadya nkhomaliro, ndipo pambali pa mnzangayu palinso Mutu wa Sayansi ya Data ndi Woyang'anira Zogulitsa yemwe ndi wokonda kwambiri Kuphunzira Kwambiri. Pa nkhomaliro tinacheza pa DL. Ndipo popeza ndakhala ndikuphunzitsa maukonde 24/7 kwa theka la chaka, ndikuwerenga mabuku a cubic mita, ndikuyendetsa ntchito pa Kaggle ndi zotsatira zomveka bwino, ndimatha kuyankhula za Kuphunzira Mwakuya kwa maola, potengera nkhani zatsopano komanso njira zothandiza .

Titatha nkhomaliro adandiyang'ana ndipo adati - zikuwonekeratu kuti ndiwe wokongola, mukufuna kulankhula nafe? Kuphatikiza apo, adawonjezeranso kuti zikuwonekeratu kuti chowonera kunyumba + chatekinoloje chikhoza kudumpha. Ndipo ndidzayitanidwa nthawi yomweyo kuti ndikafike pamalopo. Ndinavomera.

Pambuyo pake, wondilemba ntchitoyo anandiyitana kuti ndikonzekere kuyankhulana pa malo, ndipo sanakhutire. Iye anang’ung’udza chinachake chokhudza kusalumphira pamutu pako.

Anabwera. Kuyankhulana pa tsamba. Maola asanu olankhulana ndi anthu osiyanasiyana. Panalibe funso limodzi lokhudza Kuphunzira Mwakuya, kapena kuphunzira pamakina mwadongosolo. Popeza palibe Deep Learning / Computer Vision, ndiye kuti sindikufuna. Choncho, zotsatira zoyankhulana zinali orthogonal.

Wolemba ntchitoyu akuyimba ndikunena kuti - zikomo, mwafika pa kuyankhulana kwachiwiri pa tsamba. Izi zonse nzodabwitsa. Chachiwiri chili pa malopo chiyani? Sindinamvepo za chinthu choterocho. Ndinapita. Pali maola angapo kumeneko, nthawi ino zonse zokhudza kuphunzira kwamakina achikhalidwe. Ndi bwino. Koma osati chidwi.

Wolemba ntchitoyo akuyimba ndi zikomo kuti ndapambana kuyankhulana kwachitatu pamalopo ndikulonjeza kuti iyi ikhala yomaliza. Ndinapita kukawona ndipo panali DL ndi CV.

Ndinali ndi m'mbuyomu kwa miyezi yambiri yemwe anandiuza kuti sipadzakhala mwayi. Sindiphunzitsa luso laukadaulo, koma pa zofewa. Osati kumbali yofewa, koma chifukwa chakuti malowa adzatsekedwa kapena kuti kampaniyo sikugwira ntchito, koma ikungoyesa msika ndi mlingo wa ofuna.

Pakati pa August. Ndinamwa mowa bwino. Malingaliro amdima. Miyezi 8 yadutsa ndipo palibe chopereka. Ndibwino kukhala wopanga moŵa, makamaka ngati luso lake ndi lachilendo. Lingaliro limabwera m'maganizo mwanga. Ndimagawana ndi Alexey Shvets, yemwe panthawiyo anali postdoc ku MIT.

Nanga bwanji ngati mutenga msonkhano wapafupi wa DL/CV, onani mipikisano yomwe imachitika ngati gawo lake, phunzitsani kena kake ndikugonjera? Popeza akatswiri onse kumeneko akupanga ntchito zawo pa izi ndipo akhala akuchita izi kwa miyezi yambiri kapena zaka, tilibe mwayi. Koma sizowopsa. Timapanga kugonjera kopindulitsa, kuwulukira kumalo otsiriza, ndipo pambuyo pake timalemba chisanadze kapena nkhani yokhudza momwe ife sitili ngati wina aliyense ndikulankhula za chisankho chathu. Ndipo nkhaniyi ili kale pa LinkedIn komanso mukuyambiranso.

Izi zikutanthauza kuti, zikuwoneka kuti ndizofunikira ndipo pali mawu osakira olondola pakuyambiranso, zomwe ziyenera kuonjezera pang'ono mwayi wofika pazenera laukadaulo. Khodi ndi zotumiza kuchokera kwa ine, zolemba za Alexey. Masewera, ndithudi, koma bwanji?

Zosavuta kuzinena koma zovuta kuchita. Msonkhano wapafupi womwe tidachezera pa Google unali MICCAI ndipo kunali mipikisano kumeneko. Tinagunda yoyamba. Zinali KUSINTHA KWA Zithunzi Zam'mimba (GIANA). Ntchitoyi ili ndi ntchito zazing'ono zitatu. Panatsala masiku 3 kuti tsiku lomaliza lifike. Ndinadzuka m'mawa, koma sindinasiye lingalirolo. Ndinatenga mapaipi anga kuchokera ku Kaggle ndikuwasintha kuchoka pa satelayiti kupita ku azachipatala. 'fit_predict'. Alexey adakonza mafotokozedwe amasamba awiri a mayankho pavuto lililonse, ndipo tidatumiza. Okonzeka. Mwachidziwitso, mutha kutulutsa mpweya. Koma zidapezeka kuti panali ntchito ina yochitira msonkhano womwewo (Gawo la Robotic Instrument) ndi ntchito zazing'ono zitatu komanso kuti tsiku lomaliza lidakwezedwa ndi masiku 4, ndiye kuti, titha kuchita 'fit_predict' pamenepo ndikutumiza. Ndi zomwe tinachita.

Mosiyana ndi Kaggle, mipikisano iyi inali ndi mfundo zawozawo zamaphunziro:

1. Palibe Bolodi. Zopereka zimatumizidwa ndi imelo.
2. Mudzachotsedwa ngati woimira gulu sabwera kudzapereka yankho pa msonkhano pa Msonkhano.
3. Malo anu pa boardboard amadziwika panthawi ya msonkhano. Sewero la maphunziro.

Msonkhano wa MICCAI 2017 unachitikira ku Quebec City. Kunena zowona, pofika Seputembala ndidayamba kupsa mtima, motero lingaliro lopumula sabata limodzi ndikupita ku Canada lidawoneka losangalatsa.

Anabwera kumsonkhano. Ndabwera ku Workshop iyi, sindikudziwa aliyense, ndikukhala pakona. Aliyense amadziwana, amalankhulana, amataya mawu ochenjera azachipatala. Ndemanga za mpikisano woyamba. Ophunzira amalankhula ndikukambirana za zisankho zawo. Ndikozizira pamenepo, ndi kunyezimira. Nthawi yanga. Ndipo ine mwanjira ina ngakhale manyazi. Iwo anathetsa vutoli, anagwirapo ntchito, sayansi yapamwamba, ndipo ndife "oyenera_kulosera" kuchokera ku zochitika zakale, osati za sayansi, koma kuti tiwonjezere kuyambiranso kwathu.

Anatuluka n’kunena kuti inenso sindine katswiri wa zamankhwala, anapepesa chifukwa chowawonongera nthawi, ndipo anandionetsa silaidi imodzi yokhala ndi yankho lake. Ndinapita kuholo.

Amalengeza ntchito yaing'ono yoyamba - ndife oyamba, komanso mochepera.
Wachiwiri ndi wachitatu akulengezedwa.
Amalengeza chachitatu - kachiwiri koyamba ndi kutsogolera.
General ndiye woyamba.

Official press release.

Ena mwa omvera akumwetulira ndikundiyang'ana mwaulemu. Ena, omwe mwachiwonekere ankaonedwa ngati akatswiri pa ntchitoyi, adalandira ndalama zothandizira ntchitoyi ndipo akhala akuchita izi kwa zaka zambiri, anali ndi mawonekedwe opotoka pang'ono pa nkhope zawo.

Chotsatira ndi ntchito yachiwiri, yomwe ili ndi ntchito zazing'ono zitatu ndipo yapititsidwa patsogolo ndi masiku anayi.

Apa ndinapepesanso ndikuonetsanso silaidi yathu imodzi.
Nkhani yomweyi. Awiri woyamba, sekondi imodzi, wamba woyamba.

Ndikuganiza kuti iyi ndi nthawi yoyamba m'mbiri kuti bungwe lotolera ndalama lapambana mpikisano wojambula zithunzi zachipatala.

Ndipo tsopano ndikuyima pa siteji, akundipatsa diploma yamtundu wina ndipo ndikuphulitsidwa. Zingakhale bwanji zimenezo? Ophunzirawa akuwononga ndalama za okhometsa misonkho, akugwira ntchito kuti achepetse komanso kupititsa patsogolo ntchito za madokotala, ndiye kuti, mwachidziwitso, zaka za moyo wanga, ndipo bungwe lina linang'amba antchito onse a maphunziro awa ku mbendera ya ku Britain madzulo angapo.

Bhonasi pa izi ndikuti m'magulu ena, ophunzira omaliza maphunziro omwe akhala akugwira ntchito izi kwa miyezi yambiri adzakhala ndi kuyambiranso komwe kuli kosangalatsa kwa HR, ndiye kuti, amafika mosavuta pazenera laukadaulo. Ndipo pamaso panga pali imelo yolandiridwa kumene:

A Googler recently referred you for the Research Scientist, Google Brain (United States) role. We carefully reviewed your background and experience and decided not to proceed with your application at this time.

Nthawi zambiri, ndili pasiteji, ndimafunsa omvera kuti: “Kodi alipo amene akudziwa kumene ndimagwira ntchito?” Mmodzi mwa omwe adakonza mpikisanowo adadziwa - adachita Google zomwe TrueAccord inali. Ena onse sali. Ndikupitiriza kuti: “Ndimagwira ntchito ku bungwe lotolera zinthu, ndipo kuntchito sindichita Computer Vision kapena Deep Learning. Ndipo m'njira zambiri, izi zimachitika chifukwa madipatimenti a HR a Google Brain ndi Deepmind amasefa kuyambiranso kwanga, osandipatsa mwayi wowonetsa maphunziro aukadaulo. "

Iwo anapereka satifiketi, yopuma. Gulu la ophunzira limandikokera pambali. Zinapezeka kuti ili ndi gulu la Health lomwe lili ndi Deepmind. Anachita chidwi kwambiri moti nthawi yomweyo anafuna kulankhula nane za ntchito ya Research Engineer mu gulu lawo. (Tinayankhulana. Kukambitsirana kumeneku kunatha kwa miyezi 6, ndinadutsa kunyumba, mafunso, koma ndinafupikitsidwa pazithunzi zamakono. Miyezi ya 6 kuyambira chiyambi cha kuyankhulana mpaka pazithunzi zamakono ndi nthawi yayitali. Kudikira kwautali kumapereka kukoma. Research Engineer ku Deepmind ku London, kumbuyo kwa TrueAccord panali sitepe yamphamvu, koma motsutsana ndi maziko a malo anga pano ndikutsika. kuti sanatero.)

Pomaliza

Panthawi yomweyi, ndinalandira mwayi wochokera kwa Lyft, womwe ndinavomera.
Kutengera zotsatira zamipikisano iwiriyi ndi MICCAI, zotsatirazi zidasindikizidwa:

1. Kugawikana kwa zida zodziwikiratu mu opaleshoni yothandizidwa ndi loboti pogwiritsa ntchito kuphunzira mozama
2. Kuzindikira kwa angiodysplasia ndi kumasulira kwamalo pogwiritsa ntchito maukonde ozama a convolutional neural
3. Vuto la magawo a zida za Robotic 2017

Ndiko kuti, ngakhale kuti lingalirolo ndi lopanda nzeru, kuwonjezera zolemba zowonjezera ndi zolembera zampikisano zimagwira ntchito bwino. Ndipo m’zaka zotsatira tinkaipiraipira.

Ndakhala ndikugwira ntchito ku Lyft kwa zaka zingapo zapitazi ndikuchita Computer Vision/Deep Learning for Self Driving cars. Ndiko kuti, ndinapeza zomwe ndinkafuna. Ndi ntchito, ndi kampani yapamwamba, ndi ogwira nawo ntchito amphamvu, ndi zina zonse zabwino.

M'miyezi iyi, ndidalumikizana ndi makampani akulu onse a Google, Facebook, Uber, LinkedIn, komanso ndi zoyambira zazikulu zosiyanasiyana.

Zinandipweteka miyezi yonseyi. Chilengedwe chimakuuzani chinthu chosasangalatsa tsiku lililonse. Kukanidwa nthawi zonse, kulakwitsa nthawi zonse ndipo zonsezi zimakongoletsedwa ndi kusowa chiyembekezo. Palibe zitsimikizo kuti mupambana, koma pali kumverera kuti ndinu opusa. Zimandikumbutsa kwambiri momwe ndinayesera kupeza ntchito nditangomaliza yunivesite.

Ndikuganiza kuti ambiri anali kufunafuna ntchito m’chigwachi ndipo zonse zinali zosavuta kwa iwo. Chinyengo, mwa lingaliro langa, ndi ichi. Ngati mukuyang'ana ntchito m'munda womwe mukumvetsetsa, khalani ndi zokumana nazo zambiri, ndipo pitilizani kwanu kunena chimodzimodzi, palibe mavuto. Ndinaitenga ndikupeza. Pali ntchito zambiri.

Koma ngati mukufuna ntchito m'munda watsopano kwa inu, ndiye kuti, pamene palibe chidziwitso, palibe kugwirizana ndi pitilizani wanu akunena chinachake cholakwika - pa nthawi ino chirichonse chimakhala chosangalatsa kwambiri.

Pakali pano, olemba ntchito amandilembera nthawi zonse ndikudzipereka kuti achite zomwe ndikuchita tsopano, koma mu kampani ina. Yakwana nthawi yosintha ntchito. Koma palibe chifukwa choti ndichite zomwe ndidachita bwino. Zachiyani?

Koma chimene ine ndikufuna, ine kachiwiri alibe chidziwitso kapena mizere pitilizani wanga. Tiyeni tiwone momwe zonsezi zimathera. Ngati zonse zikuyenda bwino, ndilemba gawo lotsatira. 🙂

Source: www.habr.com