Vineet Sharma profile photo 1 on Tal

Vineet Sharma

MLOps and Data Engineering | IIIT

Bengaluru

5+ years of work experience

ResMed
Feb 20242 yrs 7 mosPresent

Bengaluru, Karnataka, India

● Deployed consumer facing RAG based chatbot - Dawn - in production ○ Setup design to load test 1k users requests for EKS based model API and implement load testing using locust ○ Tune parameters for API deployment for scaling pod replicas and unicorn workers for the API ○ Deployed AWS Opensearch as vector store and setup hybrid search for matching relevant docs from vector store ○ Designed and deployed scalable SLM inference APIs on EKS ○ Setup canary deployment using Argo Rollout for the API to new version release ● Setup and deploy ML Platform in AMR and EU region for nonprod and prod to host 2000+ user services ○ Setup IaC and CI/CD github actions to deploy EKS clusters per environment ○ Deployed AWS Privatelink connection between On-premise datacenter and AWS account through hybrid DNS ○ Setup and deployed using ArgoCD helm charts for Prometheus, Inferentia Operator and Nvidia GPU operator ○ Setup Kyverno policies to enforce tags on workloads running on platform ○ Debug platform and user workload issues ● Led design and deployment of EKS based ML platform monitoring layer over Datadog by managing a team of 3 engineers. ○ Setup reusable terraform module which allows creating new monitor in datadog within 15 mins ○ Setup SLO monitors for production ML workload for SLIs
Episource
Apr 202310 mosJan 2024

Mumbai, Maharashtra, India

Working in NLP & Data Science Team where my work revolves around MLOps and Platform Engineering ● Deployed model real time monitoring layer for multi-model ML system with Snowflake as data warehouse and analytics layer for 2M+ data points per hour; provide insights through SQL reporting to reduce overall pipeline time by 30% ● Deployed Kafka consumers as microserverice in EKS to get model inference in event driven pipeline ● Deployed extractive ML model on AWS Sagemaker and tuned deployment parameters to support 100k ML API request / min
Episource
Apr 20221 yrMar 2023

Mumbai, Maharashtra, India

- Event driven multi-model pipeline * Deployed SQS broker and Celery workers+infra in EKS to autoscale with SQS queue length * Deployed Airflow DAG in Kubernetes executor environment to ingest 22K messages in parallel into SQS in 5 mins - Deployed infrastructure for ML Pipeline * Setup Batch as IaC with optimal scheduling policy to run multiple batch jobs in parallel * Setup Lambda as IaC with multiple validation layers to verify and pre-process data * Setup DynamoDB as IaC with autoscaling for read/write capacity * Refactored existing workflows to build custom packer-based AMIs * Collaborated with geographically distributed team - PoC on AWS EventBridge * Setup eventbus, rules, triggers and IAM policy as IaC * Tested target trigger via put event script - PoC on AWS Opensearch cluster * Setup infrastructure as IaC with optimal storage, node and indexing strategy for Opensearch * Added script to index ~1million rows per minute * Load tested Opensearch cluster to monitor key performance indicators * Developed domain access polices - Tech Stack : Python, Celery, Pandas, Git, Terraform, API Load Testing, Airflow, CI/CD, RDBMS, SQL, AWS - Opensearch, Lambda, IAM, EKS, EventBridge, DynamoDB, Batch, CodeBuild, RDS
Episource
Jul 20219 mosMar 2022

- Developed multi-repo documentation microservice from scratch and deployed on AWS * Designed and implemented CI/CD workflows for continuous integration and validation of documentation * Researched and configured Sphinx to generate static site from MD, rST files and auto-generate API documentation from code * Provisioned infrastructure resources on AWS, using Terraform, to run docserver as containerized service on ECS * Documented the service * Generated Sphinx based documentation of few repositories and integrated with the service. - Deployed DAG on Kubernetes Executor based Airflow environment * Developed task script to query Athena * Developed task script to copy 1200 files per minute in parallel in S3 * Deployed React-Flask based dashboard to provide input to DAG via Kafka based pub/sub layer * Developed scripts to asynchronously query multiple, nested endpoints in Airflow REST API * Documented the tool with UMLs - Deployed Fluent-bit as Daemonset in EKS cluster for routing logs to CloudWatch * Documented the log agent - Developed project template for Airflow DAG to automate DAG template generation through CLI * Documented the CLI tool - Migrated apps running in ECS to EKS with argo - Tech Stack : Airflow, Kubernetes, Terraform, Docker, Kafka, Argo, GitHub Actions, Sphinx, Nginx, React, Flask, PyTest, UML Modelling, DVC, AWS - ECS, EKS, EC2-Autoscaling, Athena, S3, CloudWatch, ECR, KMS
The Solar Labs
May 20193 mosJul 2019

Noida, Uttar Pradesh, India

- Product was developed for Government of Maharashtra to analyse rooftop solar potential of the state. - Researched, designed and developed the GIS backend architecture, DB schema and APIs - Extensively used ArcGISJS API for geocoding and reverse-geocoding, pin-pointing a location, tracing an area, 3D visualization of shapefile data and to find intersecting shape-ids under an traced id - Developed a script to automate shapefile data addition into database - Developed and designed intuitive UI - Took complete ownership of the tool starting from designing from scratch to development - Tech Stack : Django, Django Rest Framework, Postgres, VueJS, UML Modelling, GIS extensions for Django, Postgres and ArcGIS
GitHub
@vineet-sharma29

Vineet’s AI Stack

github copilot
github copilot
langchain
langchain
langgraph
langgraph
notion
notion

Education

Indian Institute of Information Technology, SriCity

Bachelor of Technology - BTech, Computer Science and Engineering

Indian Institute of Information Technology, SriCity

20174 yrs2021

Join Vineet and 10k+ other professionals on

Tal Boss

Tal is the app where bosses meet engineers & builders directly, without middlemen

Download Tal Boss

Vineet Sharma on Tal

Headline

MLOps and Data Engineering | IIIT

About

MLOps / Data Platform Engineer with 5+ years of experience designing and scaling ML infrastructure and data pipelines on AWS. Expertise in building event-driven distributed systems and ML systems using Kafka, deploying containerized ML workloads on Kubernetes, and managing end-to-end ML lifecycle with Sagemaker and DVC. Strong background in infrastructure automation using Terraform and building production-grade observability systems.. Passionate about problem-solving and finding better solutions for the existing one, I am an avid team player and love contributing to impactful projects. Tech Skills Languages: Python, SQL MLOps: MLflow, Feast, DVC, AWS SageMaker, GenAI, LLM, LangGraph, LangChain Data: Airflow, Kafka, Spark, Snowflake, dbt, ETL/ELT DevOps: Docker, Kubernetes(EKS), Terraform,Helm, Github Actions, ArgoCD, SLI, SLO, SLA Observability: Prometheus, Datadog, FluentBit Databases: Postgres, DynamoDB Cloud AWS: S3, RDS, ECS, EC2, DynamoDB, OpenSearch, SQS, IAM, Lambda, Eventbridge, Lambda, CodeBuild, EventBridge Other: UML Modelling

Location

Bengaluru

AI stack

github copilot

langchain

langgraph

notion

Work experience

MLOps Engineer at ResMed, Feb 2024 - Present · 2 yrs 7 mos, Bengaluru, Karnataka, India, ● Deployed consumer facing RAG based chatbot - Dawn - in production ○ Setup design to load test 1k users requests for EKS based model API and implement load testing using locust ○ Tune parameters for API deployment for scaling pod replicas and unicorn workers for the API ○ Deployed AWS Opensearch as vector store and setup hybrid search for matching relevant docs from vector store ○ Designed and deployed scalable SLM inference APIs on EKS ○ Setup canary deployment using Argo Rollout for the API to new version release ● Setup and deploy ML Platform in AMR and EU region for nonprod and prod to host 2000+ user services ○ Setup IaC and CI/CD github actions to deploy EKS clusters per environment ○ Deployed AWS Privatelink connection between On-premise datacenter and AWS account through hybrid DNS ○ Setup and deployed using ArgoCD helm charts for Prometheus, Inferentia Operator and Nvidia GPU operator ○ Setup Kyverno policies to enforce tags on workloads running on platform ○ Debug platform and user workload issues ● Led design and deployment of EKS based ML platform monitoring layer over Datadog by managing a team of 3 engineers. ○ Setup reusable terraform module which allows creating new monitor in datadog within 15 mins ○ Setup SLO monitors for production ML workload for SLIs

Data Engineer II(MLOps+Data) at Episource, Apr 2023 - Jan 2024 · 10 mos, Mumbai, Maharashtra, India, Working in NLP & Data Science Team where my work revolves around MLOps and Platform Engineering ● Deployed model real time monitoring layer for multi-model ML system with Snowflake as data warehouse and analytics layer for 2M+ data points per hour; provide insights through SQL reporting to reduce overall pipeline time by 30% ● Deployed Kafka consumers as microserverice in EKS to get model inference in event driven pipeline ● Deployed extractive ML model on AWS Sagemaker and tuned deployment parameters to support 100k ML API request / min

Data Engineer I at Episource, Apr 2022 - Mar 2023 · 1 yr, Mumbai, Maharashtra, India, - Event driven multi-model pipeline * Deployed SQS broker and Celery workers+infra in EKS to autoscale with SQS queue length * Deployed Airflow DAG in Kubernetes executor environment to ingest 22K messages in parallel into SQS in 5 mins - Deployed infrastructure for ML Pipeline * Setup Batch as IaC with optimal scheduling policy to run multiple batch jobs in parallel * Setup Lambda as IaC with multiple validation layers to verify and pre-process data * Setup DynamoDB as IaC with autoscaling for read/write capacity * Refactored existing workflows to build custom packer-based AMIs * Collaborated with geographically distributed team - PoC on AWS EventBridge * Setup eventbus, rules, triggers and IAM policy as IaC * Tested target trigger via put event script - PoC on AWS Opensearch cluster * Setup infrastructure as IaC with optimal storage, node and indexing strategy for Opensearch * Added script to index ~1million rows per minute * Load tested Opensearch cluster to monitor key performance indicators * Developed domain access polices - Tech Stack : Python, Celery, Pandas, Git, Terraform, API Load Testing, Airflow, CI/CD, RDBMS, SQL, AWS - Opensearch, Lambda, IAM, EKS, EventBridge, DynamoDB, Batch, CodeBuild, RDS

Associate Data Engineer at Episource, Jul 2021 - Mar 2022 · 9 mos, - Developed multi-repo documentation microservice from scratch and deployed on AWS * Designed and implemented CI/CD workflows for continuous integration and validation of documentation * Researched and configured Sphinx to generate static site from MD, rST files and auto-generate API documentation from code * Provisioned infrastructure resources on AWS, using Terraform, to run docserver as containerized service on ECS * Documented the service * Generated Sphinx based documentation of few repositories and integrated with the service. - Deployed DAG on Kubernetes Executor based Airflow environment * Developed task script to query Athena * Developed task script to copy 1200 files per minute in parallel in S3 * Deployed React-Flask based dashboard to provide input to DAG via Kafka based pub/sub layer * Developed scripts to asynchronously query multiple, nested endpoints in Airflow REST API * Documented the tool with UMLs - Deployed Fluent-bit as Daemonset in EKS cluster for routing logs to CloudWatch * Documented the log agent - Developed project template for Airflow DAG to automate DAG template generation through CLI * Documented the CLI tool - Migrated apps running in ECS to EKS with argo - Tech Stack : Airflow, Kubernetes, Terraform, Docker, Kafka, Argo, GitHub Actions, Sphinx, Nginx, React, Flask, PyTest, UML Modelling, DVC, AWS - ECS, EKS, EC2-Autoscaling, Athena, S3, CloudWatch, ECR, KMS

Software Development Intern at The Solar Labs, May 2019 - Jul 2019 · 3 mos, Noida, Uttar Pradesh, India, - Product was developed for Government of Maharashtra to analyse rooftop solar potential of the state. - Researched, designed and developed the GIS backend architecture, DB schema and APIs - Extensively used ArcGISJS API for geocoding and reverse-geocoding, pin-pointing a location, tracing an area, 3D visualization of shapefile data and to find intersecting shape-ids under an traced id - Developed a script to automate shapefile data addition into database - Developed and designed intuitive UI - Took complete ownership of the tool starting from designing from scratch to development - Tech Stack : Django, Django Rest Framework, Postgres, VueJS, UML Modelling, GIS extensions for Django, Postgres and ArcGIS

Education

Bachelor of Technology - BTech, Computer Science and Engineering, Indian Institute of Information Technology, SriCity

Public links

github: https://github.com/vineet-sharma29