Snehasish Roy profile photo 1 on Tal

Snehasish Roy

Distributed Systems • Engineering at PhonePe • Ex {Google, Amazon, Goldman Sachs, Morgan Stanley} • IIIT Allahabad

Bengaluru

11+ years of work experience

PhonePe
Nov 20223 yrs 10 mosPresent

Bengaluru

Building Platform at PhonePe ! Projects which impacted the entire Company: 1. Scaled Distributed Job Scheduler Platform (Clockwork) to handle 100K RPS from 10K RPS with the same infra. Clockwork currently handles 3 Billion daily callbacks with p99 lag < 5 sec. 2. Scaled Object Storage Platform (Docstore) to handle 7 Billion Documents at 1PB Scale. Drove 30% infra cost reduction in Docstore by desiging File Packing, and FileTiering Workflow Orchestrator. To drive adoption, led S3 and SFTP Compatible interfaces. 3. Proposed and developed AtomDB (Strongly Consistent Embedded KV Database) from scratch to eliminate the operational complexity of external datastores. Know more at: Clockwork: https://tech.phonepe.com/clockwork-the-backbone-of-phonepes-2-billion-daily-jobs/ Docstore: https://archive.fosdem.org/2023/schedule/event/sds_lessons_learnt_glusterfs/ AtomDB: https://snehasishroy.com/introducing-atomdb-a-strongly-consistent-embedded-keyvalue-database-built-from-scratch Public Talks: https://www.meetup.com/open-data-platform-and-data-observability-by-acceldata/events/315265254/ https://hasgeek.com/rootconf/platform-engineering-meet-up-may-9/crew
Google
Apr 20211 yr 8 mosNov 2022

Bengaluru

1. Led the design of a MapReduce Pipeline generating daily materialized snapshots of all User Generated Content moderation decisions in Google Maps, eliminating expensive multi-join queries for downstream consumers, allowing any team to perform analysis without pipeline expertise. 2. Optimized the pipeline to ingest 40 TB of data and generate snapshots of 1 TB within 2 hours at p99.
Amazon
Apr 20192 yrs 1 moApr 2021

Bengaluru, Karnataka, India

Part of the Cart Team in Amazon 1. Reduced the latency of the cart page by 50ms @ p90 by leading the initiative to optimize product-related notifications on Cart Page. Wrote the framework for displaying new notifications in a clean and scalable manner by reducing code duplicity whilst keeping the ease of partner development in mind. 2. As part of Stretch goals, on-boarded JRebel, which allowed hot reloading of class files, saving developer time (5 min per build). 3. As part of another Stretch goal, evaluated the impact of removing white-spaces and empty new lines from the rendered HTML view, which can reduce the overall markup size by 30%, potentially giving a client-side latency win. 4. Received team award for Insisting on Highest Standards during the yearly org event. 5. Received 2 accolades from team members for Insisting on Highest Standards and Dive Deep. 6. As part of internal learning sessions, presented talks on how to be a 10x developer and shared insights on the threading architecture of the internal codebase.
Goldman Sachs
Jan 20181 yr 4 mosApr 2019

Bengaluru Area, India

Part of the Equity Risk Strat Team at Goldman Sachs 1. Reduced the latency of the Risk Infra platform used firm-wide by 5x @ p50 by spearheading the initiative to optimize the underlying data store. 2. Reduced latency by leveraging in-memory distributed database and eliminated hotspots in long-running SQL queries using Query profiling. 3. Developed a custom distributed semaphore framework to throttle memory-intensive queries. 4. Revamped the existing risk dashboard to display SLO metrics to provide better visibility to stakeholders for efficient SLA tracking. 5. Co-Mentored summer intern on Real-time SRE Pipeline goal which involved working on Apache Kafka, ElasticSearch and Kibana.
Goldman Sachs
Oct 20174 mosJan 2018

Bengaluru

Morgan Stanley
Jan 201710 mosOct 2017

Bangalore

Part of the Equity Risk Team at Morgan Stanley 1. Reduced memory utilization by 50% @ p99 of the data aggregation platform used in Risk Tech by implementing a framework that asynchronously wrote to disk while data was asynchronously read and transformed. 2. Improved throughput by 30% @ p99 by developing a smart cache solution. 3. Improved performance by analyzing GC logs to tweak JVM arguments, and leveraged JProfiler/VisualVM to identify hotspots in code. 4. Underwent four months Technical Analyst Training Program (TAP) by Mallon Associates that covered Linux Fundamental, Design Patterns, Dependency Injection, Object Oriented Approaches in C++, Java, C#. Functional Programming in Java 8, Scala, web development essentials.
Morgan Stanley
Jan 20161 yr 1 moJan 2017

Bengaluru

Morgan Stanley
Jul 20157 mosJan 2016

Mumbai

GitHub
@snehasishroy
Medium
@snehasishroy
Personal portfoliosnehasishroy.com

Snehasish’s AI Stack

claude code
claude code
kimi
kimi
langchain
langchain
langgraph
langgraph
notion
notion
Devin
Devin

Education

Indian Institute Of Information Technology allahabad

Bachelor of Technology - BTech, Information Technology

Indian Institute Of Information Technology allahabad

20114 yrs2015

Join Snehasish and 10k+ other professionals on

Tal Boss

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

Download Tal Boss

Snehasish Roy on Tal

Headline

Distributed Systems • Engineering at PhonePe • Ex {Google, Amazon, Goldman Sachs, Morgan Stanley} • IIIT Allahabad

About

Technology Leader with extensive experience in taking product, engineering, and innovation from 0 to 1 and then scaling it by 10X. Grounded in engineering fundamentals with over a decade of experience in building high-availability platforms and scalable backend systems from the ground up. My focus areas include engineering excellence and execution at scale. Passionate about driving innovation and creating systems that deliver business impact. Skilled in Java, Spring, Distributed Databases (HBase, Aerospike, MemSQL, HBase, ElasticSearch) along with MariaDB, Kafka, RabbitMQ. Completed Bachelor of Technology (BTech) with Honors focused in IT from IIIT Allahabad with a CGPI of 9.15. Competitive Programming Stats: 1. LeetCode: Max Contest Rating of 2011, with a worldwide ranking of ~7K out of ~220K users. Best ranking of 228 in LeetCode Biweekly Contest 56. Second Best ranking of 466 in LeetCode Biweekly Contest 60. Third Best ranking of 578 in LeetCode Biweekly Contest 74. 2. CodeForces: Max rating of 1423 (Specialist) 3. CodeChef: Max rating of 1665 4. GeeksForGeeks: Achieved 27 rank out of ~1200 contestants in GFG Coding Challenge https://practice.geeksforgeeks.org/contest/the-easiest-ever-coding-challenge/leaderboard/ https://drive.google.com/file/d/1YS8GoZtE2nH0dnlcGWqWnjxzZbVt1WFh/view 5. Facebook Hacker Cup 2021 Qualification Round 2021: Rank 746 Worldwide, Rank 149 India Round 1 2021: Rank 1327 Worldwide, Rank 220 India Round 2 2021: Rank 2775 Worldwide, Rank 527 India https://www.facebook.com/codingcompetitions/hacker-cup/2021/certificate/661693404384805

Location

Bengaluru

AI stack

claude code

kimi

langchain

langgraph

notion

Devin

Work experience

Technical Lead at PhonePe, Nov 2022 - Present · 3 yrs 10 mos, Bengaluru, Building Platform at PhonePe ! Projects which impacted the entire Company: 1. Scaled Distributed Job Scheduler Platform (Clockwork) to handle 100K RPS from 10K RPS with the same infra. Clockwork currently handles 3 Billion daily callbacks with p99 lag < 5 sec. 2. Scaled Object Storage Platform (Docstore) to handle 7 Billion Documents at 1PB Scale. Drove 30% infra cost reduction in Docstore by desiging File Packing, and FileTiering Workflow Orchestrator. To drive adoption, led S3 and SFTP Compatible interfaces. 3. Proposed and developed AtomDB (Strongly Consistent Embedded KV Database) from scratch to eliminate the operational complexity of external datastores. Know more at: Clockwork: https://tech.phonepe.com/clockwork-the-backbone-of-phonepes-2-billion-daily-jobs/ Docstore: https://archive.fosdem.org/2023/schedule/event/sds_lessons_learnt_glusterfs/ AtomDB: https://snehasishroy.com/introducing-atomdb-a-strongly-consistent-embedded-keyvalue-database-built-from-scratch Public Talks: https://www.meetup.com/open-data-platform-and-data-observability-by-acceldata/events/315265254/ https://hasgeek.com/rootconf/platform-engineering-meet-up-may-9/crew

SWE 3 at Google, Apr 2021 - Nov 2022 · 1 yr 8 mos, Bengaluru, 1. Led the design of a MapReduce Pipeline generating daily materialized snapshots of all User Generated Content moderation decisions in Google Maps, eliminating expensive multi-join queries for downstream consumers, allowing any team to perform analysis without pipeline expertise. 2. Optimized the pipeline to ingest 40 TB of data and generate snapshots of 1 TB within 2 hours at p99.

SDE 2 - Cart Platform Team at Amazon, Apr 2019 - Apr 2021 · 2 yrs 1 mo, Bengaluru, Karnataka, India, Part of the Cart Team in Amazon 1. Reduced the latency of the cart page by 50ms @ p90 by leading the initiative to optimize product-related notifications on Cart Page. Wrote the framework for displaying new notifications in a clean and scalable manner by reducing code duplicity whilst keeping the ease of partner development in mind. 2. As part of Stretch goals, on-boarded JRebel, which allowed hot reloading of class files, saving developer time (5 min per build). 3. As part of another Stretch goal, evaluated the impact of removing white-spaces and empty new lines from the rendered HTML view, which can reduce the overall markup size by 30%, potentially giving a client-side latency win. 4. Received team award for Insisting on Highest Standards during the yearly org event. 5. Received 2 accolades from team members for Insisting on Highest Standards and Dive Deep. 6. As part of internal learning sessions, presented talks on how to be a 10x developer and shared insights on the threading architecture of the internal codebase.

Technical Associate at Goldman Sachs, Jan 2018 - Apr 2019 · 1 yr 4 mos, Bengaluru Area, India, Part of the Equity Risk Strat Team at Goldman Sachs 1. Reduced the latency of the Risk Infra platform used firm-wide by 5x @ p50 by spearheading the initiative to optimize the underlying data store. 2. Reduced latency by leveraging in-memory distributed database and eliminated hotspots in long-running SQL queries using Query profiling. 3. Developed a custom distributed semaphore framework to throttle memory-intensive queries. 4. Revamped the existing risk dashboard to display SLO metrics to provide better visibility to stakeholders for efficient SLA tracking. 5. Co-Mentored summer intern on Real-time SRE Pipeline goal which involved working on Apache Kafka, ElasticSearch and Kibana.

Technical Analyst at Goldman Sachs, Oct 2017 - Jan 2018 · 4 mos, Bengaluru

Senior Technical Associate at Morgan Stanley, Jan 2017 - Oct 2017 · 10 mos, Bangalore, Part of the Equity Risk Team at Morgan Stanley 1. Reduced memory utilization by 50% @ p99 of the data aggregation platform used in Risk Tech by implementing a framework that asynchronously wrote to disk while data was asynchronously read and transformed. 2. Improved throughput by 30% @ p99 by developing a smart cache solution. 3. Improved performance by analyzing GC logs to tweak JVM arguments, and leveraged JProfiler/VisualVM to identify hotspots in code. 4. Underwent four months Technical Analyst Training Program (TAP) by Mallon Associates that covered Linux Fundamental, Design Patterns, Dependency Injection, Object Oriented Approaches in C++, Java, C#. Functional Programming in Java 8, Scala, web development essentials.

Technical Associate at Morgan Stanley, Jan 2016 - Jan 2017 · 1 yr 1 mo, Bengaluru

Technical Analyst at Morgan Stanley, Jul 2015 - Jan 2016 · 7 mos, Mumbai

Education

Bachelor of Technology - BTech, Information Technology, Indian Institute Of Information Technology allahabad

Public links

github: https://github.com/snehasishroy

medium: https://medium.com/@snehasishroy

snehasishroy.com: https://snehasishroy.com/