Working on projects for• Model Comparison o Developed and implemented a scalable system for comparing machine learning models, improving selection efficiency by 30% o Built data pipelines using Python and SQL ensuring clean data for model evaluation• Model Debugging o Developed and deployed docker container environments with tools for debugging machine learning models enhancing performance reliability o Created logging and monitoring systems, identifying and resolving model issues o Developed Python visualization tools for analyzing model outputs, contributing to a 20% increase in model stability o Collaborated with infrastructure teams, ensuring integration and operational efficiency of AI systems• AI Settings o Analyzed and optimized AI system configurations, enhancing performance and resource utilization by 25% o Automated hyperparameter tuning processes, identifying optimal settings for different models

AI Research and Development• Double promotion from Senior Software Engineer to Program Architect• Responsibilities same as Senior Software Engineer

AI Research and Development• Top Performer of H2 2020 in AI R&D• Part of the AI Advocates team which was regarded as the most successful team of the year in AI R&D o Involved in Evaluator (Afiniti service) machine learning model application design to make model structures extremely flexible thereby enabling Applied AI to deploy hierarchical model structures o Designed tests for QA team to fully test AI model Evaluator algorithms before every release, reducing revenue impacting errors on runtime. o Designed code to connect Python to Jira to build and deploy (with Docker), a dashboard in Python-Flask for visualizing Evaluator build timelines (feature differences, fixes and open issues) o Worked with Applied AI to decide and configure how their machine learning models should be structured for optimal performance o Redrafted Evaluator documentation to make it accessible for Applied AI, improving Evaluator deployment speed by 50%• Assigned trainer for Applied AI o Trained over 100 global Applied AI new hires on how Afiniti works, and how AI models are built through Bayesian inference, in 3-week courses with 3-4 hour sessions on alternate days o Trained non-AI (business) teams on how AI in general and in Afiniti works, making them better equipped to converse about Afiniti’s capabilities with clients

• Took over a struggling client as AI Tech Lead to bring performance from incremental revenue 2%, to 5% o Performed cluster analysis on callers across timeframes to get more stable patterns of caller behavior which increased model efficiency by 50% o Developed process in R to analyze the statistical significance of agent sales that caught a problem with the client’s sales system which was not logging sales made by agents. This R process had company-wide use o Analyzed usability, through call distributions, and made use of the high percentage of 1-1 paired ON calls for the time-series AI model to improve performance stability o Fixed errors in SQL matching queries used to build modelling data, taking matching percentage to above 90%• Built model validation process in R to measure expected hierarchical performance and output reports of over 40 time-series models in parallel• Responsible for calculating model performance through A/B testing to decide when to update or replace model for better/consistent CRO• Assigned Grafana reports’ lead for Applied AI o Built a fully custom Grafana dashboard for AT&T to monitor over 1 million daily calls spread over 40 call centre queues in the US o Involved in designing user-friendly interactive dashboards for company-wide model and runtime data diagnostics o Recommended company-wide database standardization to make Grafana dashboard plug-and-play which sped up their deployment time across clients from days to hours o Training AI team on how to build custom Grafana dashboards

• Top Performer of Q4 2018 in AI Production• Assigned AI Tech Lead in March 2017• Maintained monthly 3% incremental target revenue for the client, generating upwards of $1M• Led a successful launch (business understanding, data analysis, optimization criteria selection, feature engineering, model design strategy and deploying time-series models) of Afiniti on a client that delivered over $500,000 in 6 months through machine learning models for call pairing• Responsible for designing, deploying, and debugging highly-optimized MySQL queries for AI sensors to monitor all clients in Afiniti for pro-active runtime performance degradation detection of AI models• Wrote SQL data pipelines for time-based matching of calls and sales (including other data) to achieve over 95% join accuracy• Generated automated Talend data pipelines to source data from clients, process the data and load into Afiniti databases for model readiness, saving hours of workday time• Built Talend data pipelines to automate SQL hourly/daily/monthly procedures for calculating and mailing model performance results to all stakeholders• Built Markdown model reports in R for AI management to drill down or monitor runtime model performance