As I begin my new role as President of the Unified Cloud Business Unit at Rakuten Symphony, following the acquisition of Robin.io, I want to invite all the dreamers out there to share in this journey of exciting career growth, with the added advantage of leaving your mark in the industry with your work. We are doing some cutting-edge work in the cloud-native space on a large scale. We're hiring across all departments -- Engineering, Solution Architects, SRE, Sales (both Enterprise and Telco), Marketing, and Product Management. Send your resumes to partha@robin.io or jobs@robin.io.We developed a unique culture at Robin.io that we bring to the Unified Cloud Business Unit. Our goal is to develop talent so that they can achieve their full potential. Therefore, our employees have seen their life's work have the greatest impact. We have been able to accomplish more with less (at Robin, we demonstrated the highest $ARR/headcount among our industry peers). Our engineering ranks have experienced lower attrition than the industry average. A high level of innovation (> 75 patents, the most among our peers in the same industry sub-segment). We created and nurtured a very unique culture that is focused on velocity of action, a growth mindset, end-to-end thinking and personal empowerment without imposing any restrictions on staying within one's assigned swim lane. As a result, our employees' career growth has been the fastest and most rewarding, both in terms of their learning and their financial rewards. If you would like to work in such an environment, then the Unified Cloud Business Unit at Rakuten Symphony is the right place for you. We are looking for dreamers who are looking for a chance to leave their mark on the industry. Send us your resume to get started -- partha@robin.io or jobs@robin.io

BUILT: Scale Out Storage product; ADVISED: various product lines (including Information Fabric, Backup Appliance, Copy-data-management product line, Data Insight, Clustered NAS.Conceived, architected and led a team to build Veritas' first Scale Out File/Object Storage product line. After conceiving the idea, I successfully pitched to the EVPs and President of Veritas to raise funding for this project. Then architected the product and built a team to deliver on the vision. We delivered a highly functional beta product to customers within 13 months -- a record for a scale-out storage product. Features delivered within 13 months include -- (a) high performance IO (line speed 10GiGE), (b) high scalability (stored 10+ billion objects in single namespace), support for 1000s of nodes. (c) True multi-protocol support (simultaneous IO via both POSIX and Object protocols to the same file – first in the industry), (d) Replication and erasure coding based protection, (e) Policy based data management, (f) COW based file versioning (g) Stellar Usability (beta customer feedback on usability was outstanding), (i) built-in reports and (j) Support for POSIX, S3, SWIFT and SMB protocols out of the box (simultaneously on the same files), (k) Multi datacenter supportMy role: (a) conceived the idea, (b) raised funding, (c) architected core components, (d) formed team and oversaw execution, (e) wrote major portions (40%) of the codebase (including metadata subsystem, cluster configuration, IO subsystem, distributed storage protocol), (f) interfaced with beta customers, (g) presented internally (VPs, EVPs and President) and external (customers)

As the chief-architect and Engineering lead I designed and implemented a highly optimized and scalable/distributed temporal database that enables Storage and Security analytics use cases. This product is used on a daily basis by top tier marquee customers to store the metadata of billions of objects and billions of time-series data. Several optimizations have been done to allow queries to be returned in milliseconds even for large indices. The index itself is designed to be horizontally scalable and a map-reduce programming interface is implemented to run queries and consolidate results from across several nodes.In addition to architecting the product I also led the US Engineering team. We executed this project in a record time frame -- first code checkin to PO in 9 months.

As CTO, I designed and implemented a scalable distributed file system protocol that has interesting scalability features. The protocol allows for the construction of petabyte scale distributed file systems on top of commodity hardware. The file system is intended for web scale use and therefore, is designed to support concurrent I/O from thousands of users. I also implemented a content-aware storage engine for this file system. As part of this work I also implemented a massively scalable distributed RPC layer; and a space-efficient data encoding format called eXtensible Data Markup.

At Kazeon I was responsible for leading the clustering and systems projects. I architected a distributed group communication framework used in Kazeon's Information Server product. As part of this framework, I implemented a zero-copy message transmission algorithm to improve performance in the group communication protocol. Being part of the early team at Kazeon I have worked closely with several Fortune 500 accounts in deploying Kazeon's products in their data centers. Was also instrumental in salvaging several accounts by diagnosing and fixing critical issues.

Built an in-kernel layer 3 to layer 7 Network Load Balancer and Application Traffic Manager for VERITAS Cluster Server. I was the chief architect of this product and have taken it from concept stages in early 2000 to 3 major product releases and 2 minor (patch) releases in 5-½ years. The network module is plumbed in the Network Stack immediately above the Link Layer (Ethernet) and below the Network Layer (IP). TCP/IP and UDP/IP packets are received at wire-speed; relevant information from their headers and payloads is parsed and modified using user configured policies; the modified packets are then routed to appropriate backend servers. I have worked on intricate protocol level issues like TCP sequence number manipulations (Connection Splicing), Packet rewriting and Incremental Checksum Updates (or taking advantage of hardware checksum offloading when available), TCP window size adjustments, efficient IP packet fragmentation and reassembly, TCP segment re-ordering and duplicate elimination using the Sliding window algorithm, Direct Server Return (i.e, distributed sequence space manipulation to significantly improve I/O throughput) etc.Invented a novel technique for migrating live TCP socket state from one node to another. The goal is to facilitate stateful process migration from a primary node to a secondary when the primary faults (panics or is powered cycled). This patented algorithm can be configured to achieve this using a single packet copy and with minimal performance degradation. TCP state migration involves sequence number mapping, window size adjustments, wire-speed record-replay, identifying pending network I/O within the Network Stack and recreating it after failure.