Role Proficiency:

Under guidance from Senior ML Engineers develop ML models that provides accurate results with controls to solve the business problem identified using state of art techniques.

Outcomes:

      Executes relevant data wrangling activities related to the problem in order to create dataset       Conduct ML experiments to understand the feasibility and build baseline models to solve the business problem       Fine tune the baseline model for optimum performance       Test Models internally per acceptance criteria from business       Document relevant Artefacts for communicating with the business       Work with data scientists to deploy the models.       Work with product teams in planning and execution of new product releases.       Set OKRs and success steps for self/ team and provide feedback to goals for team members       Work with cross functional teams - business technology and product teams to understand the product vision; building ML solutions that provides value to the product

Measures of Outcomes:

      Selection of the appropriate approach to the problem       Number of successful deployments of the model with optimised accuracy for baseline model       Adherence to project schedule / timelines       Personal and team achievement of quarterly/yearly objectives (OKR Assignments HIG Stretch goals)

Outputs Expected:

Design to deliver Product Objectives:

Design ML solutions which are aligned to and achieve product objectives Define data requirements for the model building and model monitoring; working with product managers to get necessary data

Updated on state of art techniques in the area of AI / ML :

Perform necessary research using the latest and state of art techniques to design scalable approaches Explain the relevance of the technologies
its pros and cons to the product team; enabling accurate design experiences

Skill Examples:

     Technically strong with the ability to connect the dots      Ability to communicate the relevance of technology to the stakeholders in a simple relatable language      Curiosity to learn more about new business domains and Technology Innovation      An empathetic listener who can give and receive honest thoughtful feedback

Knowledge Examples:

      Expertise in machine learning model building lifecycle       Clear understanding of various ML techniques with appropriate use to business problems       A strong background of statistics and Mathematics       Expertise in one of the domains – Computer Vision Language Understanding or structured data       Experience in executing collaboratively with engineering design user research teams and business stakeholders       Experience with data wrangling techniques preprocessing and post processing requirements for ML solutions       Good knowledge python and deep learning frameworks like Tensorflow Pytorch Caffe       Familiar with the machine learning model testing approaches       A genuine eagerness to work and learn from a diverse and talented team

Additional Comments:

Who we are? You will be part of the ML R&D team which works on some really cool problems and (sometimes not-so-cool :-) problems). We apply cutting edge ML to solve hard problems like Document Understanding (or Document Al). We have a solution in production which is on par with the industry players in multiple facets. We reason things from the 1 st principles, or we build on top of existing things as the problem dictates. We as a team push the boundary of ML and constantly work on techniques to solve problems with no or little training data. We are a very flat org; everyone is technically sound and very collaborative. Your typical day would involve creating datasets from the scratch or run multiple iterations of feature engineering or come up with a great representation learning technique or conceptualize a nifty transfer learning solution, fit a model to the data and package the model to serve in batch or in online fashion. Who we are looking for? • We are flexible and are looking for the top talent ideally with 6+ years industry experience or 1-2 years academic experience. • Applied ML Experience: o Problem framing: ■ Strong problem framing skills: Say, when to go with Supervised or self-supervised or RL setting. o Data wrangling skills: ■ Experience in techniques like Weak/Distant Supervision and Pseudo labelling) ■ Strong EDA, data preparation and labelling skills ■ Strong data augmentation skills o From the scratch learning: ■ Strong experience in end to end modelling in (ML vs DL vs RL), ■ Experience in Single models vs Ensembles vs Mixture of experts. ■ Mathematical understanding of some Mathematical Induction, Tree Induction, DL and other optimization algorithms like SGD. o Transfer Learning ■ Experience in N-shot learning (or its variants) ■ Fine tuning skills UST Global Ltd 1 SmartOps Strategic R&D o ML/DL Verticals: Proven research or industry experience in one of the areas like Time series modelling, Vision, NLP, RL. • A GitHub portfolio with original ML repos. • A Kaggle portfolio with decent leader board positions • Papers: Original 1 st author papers in reputed ML journals or conferences. • Patents: Al or Automation specific patent is a good to have • Experience with ML/DL libraries TensorFlow or PyTorch • MLOps: Experience in running machine learning experiments with any one of the above machine learning libraries. Good to have is any one of the following: Kubeflow, Mlflow or Airflow or SparkML • Deploying machine learning solutions into production. Model Serving TFServe, Seldon, Custom serving. Interactive, batching and streamed serving. • Optimizing solutions for performance and scalability. • Data engineering, i.e. ensuring a good data flow between database and backend systems. • Implementing custom machine learning code (like custom implementation of existing algorithms like SGD) when required • Coming up with our own DNN architectures when required • Good to have: Computer science or IT background • Good to have: Exposure to statistics and probability. • Good to have: Experience in running dockerized code, we are a Kubernetes shop