Machine Learning DevOps Engineer (Pytorch, NVTabular, AWS)
Budget: $750 – $1,500 USD
As a Machine Learning Devops Engineer you’ll be running / executing deep learning models in scale using Pytorch, NVTabular & Ray within AWS.
Join an experienced data science team, working with Deep Learning and advanced Bayesian modeling approaches.
It's a super interesting project with challenging research and productization aspects, for a major US client (fortune 500), flexible working hours.
It's a great opportunity for a devops engineer who wants to learn the inside outs of deep learning and bayesian modelling approaches.
Looking for an experienced Machine Learning DevOps Engineer, with hands on experience writing python backend code through various scenarios, support the improvement of the prediction accuracy of our engine.
Responsibilities:
1. Design of evaluation pipelines:
Design and implement monitoring processes and scripts to test ML prediction accuracy
2. Load data and run:
Write SQL queries (or Python) to load relevant data into monitoring processes and run the models within Pytorch / Ray using NVTabular to showcase model accuracies results.
3. Visualize:
Design and implement time series visualizations to communicate accuracy results across multiple data segmentations.
4. Communicate:
Communicate status and recommendations effectively and proactively, both internally and externally with enterprise level engineers and executives (both internally and client side)
Must have requirements:
1. Hands on experience with Pytorch (or at least 3 years of TensorFlow) - must.
2. Hands on experience working on AWS infrastructure - must.
3. Python programmer: at least 3 years of experience
4. Visualization library: at least 2 years of experience
5. Experience deploying code to production environments regularly. Comfortable with *nix, git.
6. Communication: Fluent in English. Be able to articulate requirements and make proactive recommendations internally with the data science team and externally with client teams. Experience creating visualizations in the context of ML and/or statistical inference.
Advantages:
1. NVTabular - big advantage.
2. Ray: advantage
3. PySpark / EMR: Advantage
4. Analyst mentality: get excited about understanding the meaning, mapping the tables, and loading data from complex databases to simplify our machine learning engine data processes.
Example day for a Machine Learning DevOps Engineer:
1. Share results of latest models accuracies with the team, and highlight implications for potential ML roadmap directions
2. Get new requirements for a KPI monitoring / evaluation scripts: E.g. plot the prediction accuracy for demographics x, y, interested in products m,n
3. Write the script with Pytorch and AWS environments
4. Run our predictive models on specific subsets of the data to collect performance metrics.
5. Examine and analyze the KPI being monitored and models accuracies results
6. Plot the results via visualization plots
Admin:
You should be available to work on EST zone.
Hopefully for a long term (6 months plus) basis.
Join an experienced data science team, working with Deep Learning and advanced Bayesian modeling approaches.
It's a super interesting project with challenging research and productization aspects, for a major US client (fortune 500), flexible working hours.
It's a great opportunity for a devops engineer who wants to learn the inside outs of deep learning and bayesian modelling approaches.
Looking for an experienced Machine Learning DevOps Engineer, with hands on experience writing python backend code through various scenarios, support the improvement of the prediction accuracy of our engine.
Responsibilities:
1. Design of evaluation pipelines:
Design and implement monitoring processes and scripts to test ML prediction accuracy
2. Load data and run:
Write SQL queries (or Python) to load relevant data into monitoring processes and run the models within Pytorch / Ray using NVTabular to showcase model accuracies results.
3. Visualize:
Design and implement time series visualizations to communicate accuracy results across multiple data segmentations.
4. Communicate:
Communicate status and recommendations effectively and proactively, both internally and externally with enterprise level engineers and executives (both internally and client side)
Must have requirements:
1. Hands on experience with Pytorch (or at least 3 years of TensorFlow) - must.
2. Hands on experience working on AWS infrastructure - must.
3. Python programmer: at least 3 years of experience
4. Visualization library: at least 2 years of experience
5. Experience deploying code to production environments regularly. Comfortable with *nix, git.
6. Communication: Fluent in English. Be able to articulate requirements and make proactive recommendations internally with the data science team and externally with client teams. Experience creating visualizations in the context of ML and/or statistical inference.
Advantages:
1. NVTabular - big advantage.
2. Ray: advantage
3. PySpark / EMR: Advantage
4. Analyst mentality: get excited about understanding the meaning, mapping the tables, and loading data from complex databases to simplify our machine learning engine data processes.
Example day for a Machine Learning DevOps Engineer:
1. Share results of latest models accuracies with the team, and highlight implications for potential ML roadmap directions
2. Get new requirements for a KPI monitoring / evaluation scripts: E.g. plot the prediction accuracy for demographics x, y, interested in products m,n
3. Write the script with Pytorch and AWS environments
4. Run our predictive models on specific subsets of the data to collect performance metrics.
5. Examine and analyze the KPI being monitored and models accuracies results
6. Plot the results via visualization plots
Admin:
You should be available to work on EST zone.
Hopefully for a long term (6 months plus) basis.