Experience

ConcertAI

• United States
• Biotechnology Research
• 700 & Above Employee

• Software Engineer

  • Oct 2022 - Present



Rephrase.ai

• United States
• Online Audio and Video Media
• 1 - 100 Employee

• Software Engineer

  • Oct 2021 - Aug 2022

  Made MVP text‑to‑video app, producing continuous video output of any text transcript, in real-time ‑ a 25x speedup from existing platform. - Implemented end‑to‑end streaming pipeline using Python multiprocessing, to run on a single powerful machine with parallel stages: obtaining audio binary, extracting speech features, video frame generation, and video streaming - Found performance bottlenecks by benchmarking all stages. Maximized throughput of pipeline and GPU, by using UNIX… Show more Made MVP text‑to‑video app, producing continuous video output of any text transcript, in real-time ‑ a 25x speedup from existing platform. - Implemented end‑to‑end streaming pipeline using Python multiprocessing, to run on a single powerful machine with parallel stages: obtaining audio binary, extracting speech features, video frame generation, and video streaming - Found performance bottlenecks by benchmarking all stages. Maximized throughput of pipeline and GPU, by using UNIX pipes and RAM for data sharing, minimizing disk access - Created web‑app and REST APIs for input; integrated webRTC, HLS and RTMP protocols as choices for video output Also implemented minor features and bug‑fixes for APIs, monitoring and orchestration of the existing text‑to‑video platform, using Django, Celery and Google Cloud. Show less Made MVP text‑to‑video app, producing continuous video output of any text transcript, in real-time ‑ a 25x speedup from existing platform. - Implemented end‑to‑end streaming pipeline using Python multiprocessing, to run on a single powerful machine with parallel stages: obtaining audio binary, extracting speech features, video frame generation, and video streaming - Found performance bottlenecks by benchmarking all stages. Maximized throughput of pipeline and GPU, by using UNIX… Show more Made MVP text‑to‑video app, producing continuous video output of any text transcript, in real-time ‑ a 25x speedup from existing platform. - Implemented end‑to‑end streaming pipeline using Python multiprocessing, to run on a single powerful machine with parallel stages: obtaining audio binary, extracting speech features, video frame generation, and video streaming - Found performance bottlenecks by benchmarking all stages. Maximized throughput of pipeline and GPU, by using UNIX pipes and RAM for data sharing, minimizing disk access - Created web‑app and REST APIs for input; integrated webRTC, HLS and RTMP protocols as choices for video output Also implemented minor features and bug‑fixes for APIs, monitoring and orchestration of the existing text‑to‑video platform, using Django, Celery and Google Cloud. Show less



Amazon

• United States
• Software Development
• 700 & Above Employee

• Software Developer

  • Jul 2020 - Sep 2021

  Worked on internal Cloud platform managing infrastructure of 3000+ Amazon services: - Calculated hardware requirements and scheduled infrastructure scaling in advance of Amazon peak days ‑ eliminating business‑impacting errors and reducing manual hours - Redesigned the notification mechanism; added option of sending secure e‑mail instead of non‑actionable tickets, to reduce KPI of tenants - Fetch and display DNS endpoints on migration platform; to migrate all traffic (1Mil+… Show more Worked on internal Cloud platform managing infrastructure of 3000+ Amazon services: - Calculated hardware requirements and scheduled infrastructure scaling in advance of Amazon peak days ‑ eliminating business‑impacting errors and reducing manual hours - Redesigned the notification mechanism; added option of sending secure e‑mail instead of non‑actionable tickets, to reduce KPI of tenants - Fetch and display DNS endpoints on migration platform; to migrate all traffic (1Mil+ transactions/sec) from external load‑balancers to Native AWS infrastructure; saving millions of dollars - Onboarded all new infrastructure with correct permissions to new, Amazon‑wide security protocol. Also worked with security team to help migrate all existing permissions on infrastructure of 2K+ services. Show less Worked on internal Cloud platform managing infrastructure of 3000+ Amazon services: - Calculated hardware requirements and scheduled infrastructure scaling in advance of Amazon peak days ‑ eliminating business‑impacting errors and reducing manual hours - Redesigned the notification mechanism; added option of sending secure e‑mail instead of non‑actionable tickets, to reduce KPI of tenants - Fetch and display DNS endpoints on migration platform; to migrate all traffic (1Mil+… Show more Worked on internal Cloud platform managing infrastructure of 3000+ Amazon services: - Calculated hardware requirements and scheduled infrastructure scaling in advance of Amazon peak days ‑ eliminating business‑impacting errors and reducing manual hours - Redesigned the notification mechanism; added option of sending secure e‑mail instead of non‑actionable tickets, to reduce KPI of tenants - Fetch and display DNS endpoints on migration platform; to migrate all traffic (1Mil+ transactions/sec) from external load‑balancers to Native AWS infrastructure; saving millions of dollars - Onboarded all new infrastructure with correct permissions to new, Amazon‑wide security protocol. Also worked with security team to help migrate all existing permissions on infrastructure of 2K+ services. Show less



Amazon

• United States
• Software Development
• 700 & Above Employee

• Software Development Intern

  • May 2019 - Jul 2019

  - Wrote feature to define custom metrics for alarms, for monitoring 100+ internal services - Learnt Agile practices, CI/CD, orchestration, design patterns, creating HLD and LLD, testing and documentation - Wrote feature to define custom metrics for alarms, for monitoring 100+ internal services - Learnt Agile practices, CI/CD, orchestration, design patterns, creating HLD and LLD, testing and documentation