Machine Learning Engineer (Generative AI) @ ADF Data Science
Jun 2022 – Present
— Exploring various llm models like llama, mistral, mixtral, dolly, etc for different use cases. Also exploring best way to do fine-tuning, inference.
— Created Knowledge management system by training and fine-tuning llama2 model.
— Automating model development, validation and monitoring document generation with LLM.
— Classifying communication between agents and clients into potential and non potential complaints using LLM.
— Building tools and infrastructure to support scalable, time and cost efficient distributed modeling architecture with tools like Jenkins, Ray, AWS, EKS, Flyte, Spark.
— Worked on affiliate models to attract the right customers. Trained multiple models in a distributed environment, logged experiments through mlflow and scheduled periodic retraining pipeline through jenkins.
— Worked on Net-response model to target the right customers. Reduce training time by using spark to collect and process data from redshift and by using data parallelism based model training with ray.
— Created a low latency, scalable, dockerized and fault tolerant model deployment server on aws with ecs, alb, ecr, golang, fast api, redis, cloudwatch, etc.
— Working on internal ML library to handle our custom needs.
— Guiding a team of ML engineers and helping data scientists.
— Automating various tasks like retraining, prediction, data updation with data pipeline, model pipeline.
