Aller au contenuAller au pied de page
  • Emplois
  • Entreprises
  • Salaires
  • Pour les employeurs

      Boostez votre carrière

      Découvrez votre salaire potentiel, décrochez des emplois de rêve et partagez vos témoignages de manière anonyme.

      employer cover photo
      employer logo
      employer logo

      ONYX InSight

      Est-ce votre entreprise ?

      À propos
      Avis
      Salaires et avantages
      Emplois
      Entretiens
      Entretiens
      Recherches associées: Avis sur ONYX InSight | Offres d’emploi chez ONYX InSight | Salaires chez ONYX InSight | Avantages sociaux chez ONYX InSight
      Entretiens chez ONYX InSightEntretiens d’embauche pour Machine Learning Engineer chez ONYX InSightEntretien chez ONYX InSight


      Glassdoor

      • À propos
      • Récompenses
      • Blog
      • Nous contacter
      • Guides

      Employeurs

      • Compte employeur gratuit
      • Centre employeur
      • Blog pour les employeurs

      Informations

      • Aide
      • Règles de la communauté
      • Conditions d'utilisation
      • Confidentialité et choix publicitaires
      • Ne pas vendre ni partager mes informations
      • Outil de consentement aux cookies

      Travailler avec nous

      • Annonceurs
      • Carrières
      Télécharger l'application

      • Parcourir par :
      • Entreprises
      • Emplois
      • Lieux

      Copyright © 2008-2026. Glassdoor LLC. « Glassdoor », son logo, « Worklife Pro » et « Bowls » sont des marques déposées de Glassdoor LLC.

      Entreprises suivies

      Tenez-vous au courant des dernières opportunités et profitez de conseils d’initiés en suivant les entreprises de vos rêves.

      Recherche d’emplois

      Obtenez des recommandations et des mises à jour personnalisées en démarrant vos recherches.

      Entretien pour Machine Learning Engineer

      26 mars 2019
      Candidat à l'entretien anonyme
      Boulder, CO
      Aucune offre
      Expérience négative
      Entretien difficile

      Candidature

      J'ai postulé en ligne. Le processus a pris 2 mois. J'ai passé un entretien chez ONYX InSight (Boulder, CO) en mars 2019

      Entretien

      I applied around January 25th, 2019 via LinkedIn. Then on February 20th the recruiter reached out via e-mail where he sent me a "knowledge quiz" consisting of 3 parts: 1. a general machine learning part with 6 questions, 2. a predictive maintenance section requiring some in-depth knowledge in mechanical engineering (the posting said nothing about needing to have pre-requisite knowledge in this area) with 7 questions and 3. a machine learning task which I needed to develop. I was given 3 days to complete the "knowledge quiz" (2 of those days were weekdays). Item 1 was straightforward, but items 2 and 3 were far more involved. I had to do quite a bit of research to arrive at answers for item 2. Item 3 was a challenging, long but fun exercise. Completing all 3 items in time was a challenge given the fact that I have a full-time job. After handing in the "knowledge quiz", the recruiter contacted me back on March 6th stating that they were "impressed" with my results and invited me for a Skype interview on March 12th. They then rescheduled the interview twice and they ended up wanting an in-person interview on March 14th. The in-person interview was a 2.5 hour long affair where the interviewers arrived late to the conference room booked for the interview. They also seemed somewhat unprepared. For example, it was supposed to be a conference call with people from all around the world, but the laptop they were connecting through ran out of battery, and had to rush to get a cable for it and connect it back. One of the interviewers kept asking me questions specifically around my PhD (which had nothing to do with the role) in a somewhat rude and grilling manner (he had to leave mid-interview to complete other tasks he had to do). The recruiter was present in the interview but did not participate at all. The bulk of technical questions relevant to the position were made by another remote employee, who asked me scattered questions about my exercise and about general machine learning concepts. If I understood correctly, I did well on this interview, but their main concern was my level of experience for the role, and so asked me in an ambiguous way if I could follow up on my already long machine learning exercise (i.e. try other models and compare with my initial solution), to which I said I could follow up if asked. I then asked questions I had about the company which they answered in a satisfactory way. All of the interviewers except one then started disconnecting/leaving. The remaining interviewer and myself pleasantly chatted for a bit and he told me that he thought that I "would be a very good fit for the role" and showed me around the office, introducing me to several employees and having chats with each. I wrote them an e-mail asking when they wanted me to hand in the follow-up exercise. The aforementioned remote employee answered "How long you think it will take to build a reasonably accurate model?", so I told him I would work on that model over the weekend (it was Thursday night) and hand it in on Monday the 18th of March, as I would be very busy the following week to be able to work on anything other than my job. I then managed to have a functional model ready on Monday the 18th of March and sent it to them with annotations of things to improve and a discussion. On the early morning of Monday the 25th of March I finally got an answer saying that they would be moving forward with another candidate. Because of the level of effort and flexibility I put into this interview, I respectfully replied to get feedback as to what I could have done better. It is still soon after the fact, but I have not yet received a response.

      Questions d'entretien [14]

      Question 1

      Explain some common machine learning concepts (precision, recall, hyperparameter optimization, etc).
      Répondre à cette question

      Question 2

      List error metrics to evaluate a binary classifier.
      Répondre à cette question

      Question 3

      Explain a few methods to handle an imbalanced dataset.
      Répondre à cette question

      Question 4

      Explain how to handle missing or corrupted data in a dataset.
      Répondre à cette question

      Question 5

      Do you have experience with Spark or big data tools for machine learning?
      Répondre à cette question

      Question 6

      Please explain your most successful (first-hand) use cases of machine learning models.
      Répondre à cette question

      Question 7

      For data acquisition, how should sample time and sample rate be treated when collecting vibration data from gears and roller bearings?
      Répondre à cette question

      Question 8

      Describe some reprocessing techniques that could be used for analyzing raw vibration data from variable speed rotating machinery.
      Répondre à cette question

      Question 9

      Describe the spectral characteristics of a bearing fault versus a gear fault.
      Répondre à cette question

      Question 10

      What would be some characteristics to describe a bearing or gear as having a more severe, progressed fault by only looking at the vibration signature?
      Répondre à cette question

      Question 11

      Describe a few ways in which a machine’s power data could be used to assess its health.
      Répondre à cette question

      Question 12

      Describe a few ways to statistically evaluate temperature data from a fleet of machines.
      Répondre à cette question

      Question 13

      SCADA uses alarms and fault codes to notify the operator of the status of a machine. How could these alarms be used for reliability analysis? How could these alarms be used for performance analysis?
      Répondre à cette question

      Question 14

      During the historical data collection period (of some given time series data about wind turbine bearing temperature), several wind turbines had generator bearing problems. Four wind turbines had generator bearing failures and replacements. The symptom of bearing fault is rising temperature beyond normal range. The task is to build the ML model to detect anomaly in generator bearing and identify wind turbines that shows generator bearing defect. You are required to submit the following: List of WT’s that are suspected to have a generator bearing defect during the data period including 4 that had change out, Result showing the reason for diagnosis, and the code associated with the aforementioned. You are then given ids for wind turbines that had NO generator bearing defect (healthy).
      Répondre à cette question
      4
      avatar
      Réponse de ONYX InSight
      6y
      Our employer brand is important to us and we set a high standard in the area of candidate experience. Part of this high standard is providing feedback to applicants who invest their time in the process of applying for a position with us. As a global SME in the EngTech space we are growing fast and experience a high level of interest in our career opportunities. I’m afraid that sometimes this means our limited resources are stretched and we fail to achieve our own high standards. Clearly on this occasion we fell short. We greatly value and acknowledge your comments. Please contact me on steven.mulholland@onyxinsight.com and I will be happy to arrange an opportunity to speak with the hiring team and provide the feedback that is part of our high standard around candidate experience.

      Meilleures entreprises pour « Rémunération et avantages » près de chez vous

      avatar
      ExxonMobil
      3.6★Rémunération et avantages
      avatar
      Shell
      3.9★Rémunération et avantages
      avatar
      Schneider Electric
      4.0★Rémunération et avantages
      avatar
      Chevron
      3.8★Rémunération et avantages