Skills

  • Languages : Golang, Python, C++
  • Frameworks and tools : Kubernetes, Docker, Jenkins, Github Actions, Microservices Patterns, REST APIs, Websockets, GCP Services, AWS Services, Kafka, Prometheus, Grafana, Elasticsearch, InfluxDB, Telegraf, Slack, PagerDuty, Jira, ServiceNow
  • LLM : OpenAI, Llama3, Huggingface, vLLM, ChatGPT, fine-tuning, LoRA, LangChain

Experience

Selector AI

ETL Team Lead • Nov, 2021 — Present

  • As the ETL team lead, I designed and developed an ingestion pipeline for the Selector ecosystem. This involved tasks like collecting, filtering, extracting, transforming, and storing data in databases.
  • I created 10+ agents to collect scaled data from various sources. These engines can run natively or remotely on customer premises. Additionally, I wrote a code generation tool that automatically generates framework code. This tool saved developers time and sped up development by contributing to 70% of the agent code.
  • I also designed an agents manager that distributes workload and ensures fault tolerance (Patent Pending: Policy based distribution for SNMP collector).
  • I was the primary point of contact for customer data onboarding, investigating scale, and fixing scale issues. I successfully onboarded 20+ customers.
  • Boosted system performance within the ETL Stack through improvements. Few scale number from a customer setup: processing 80k OID’s/min from 24k devices for polling, real-time processing of ingests data at a rate of 200Mb/sec.
  • In the last 3 years, I led a team of 5 and mentored 9+. I contributed over 500K lines of code and filed 2 patents.
  • ML Infra Team Lead • Apr, 2024 — Present

  • I led a team of 3 engineers for the co-pilot project. Our goal was to build an infrastructure for creating datasets, fine-tuning LLM using LoRA, and deploying inference serving.
  • The end-to-end pipeline can train LLM and deploy it in production in less than 30 minutes.
  • The current fine-tuned model performs better than the OpenAI Berkeley function calling benchmark result.
  • We also developed a system, understanding the underlying system, to automatically generate datasets from the metrics database.
  • Uber

    Senior Software Engineer • Apr, 2019 — Nov, 2021

  • As lead for the Uber network infrastructure team, I oversee the design and development of a system that manages over 9k+ devices across eight data centers and cloud services. Network infrastructure management encompasses configuration, monitoring, remediation, and automated workflow.

  • Juniper Networks

    Software Engineer • Jul, 2013 — Apr, 2019

  • Building network infrastructure for a range of container environments. Container networking, SDN, VIP, load balancing and security are among the projects I've worked on.

  • Education

    University of Southern California

    Master of Science in Computer Science, 2011 — 2013