SigLens helps save 90% of your observability costsWHAT DO CURRENT SOLUTIONS DO?To reduce observability costs current solutions do things as follows:1) Reduce retention.This gives immediate savings but it reduces the visibility and historical comparisons2) Dump majority of it in S3 and only keep important services logs in the observability systemThis also gives good savings. However it is a risky. During production outages, the logs from S3 are pulled into the observability system. This takes time and causes prolonged outages3) Segment logs into hot, warm and cold and use different system for each segment.This is fair enough approach. By segmenting necessarily different systems are used for different use cases. It does help in reducing the cost. The downside however is maintaining and building such systems. Quite often the burden is passed on to the users to rememeber to go to different systems for different segments.HOW IS SIGLENS DIFFERENT?SigLens has build a observability database from scratch that is 100x more efficient and several magnitudes faster than current solutions in the industry. With SigLens, companies can use one single system for all data types, reduce costs by 90% and also have lightening fast query response times. SigLens helps reduce your downtime.Below are 3 key innovations with SigLens Database:1) Dynamic columnar compressions:SigLens chooses a compression compression on a column based on the past 5 minutes of data and keeps adaptively changing it as the shape of the data changes. This helps achieve more than 95% storage savings with zero configurations.2) Dynamic micro-indexingMicro-index are like the database indexes, except they are 1/100th the size of an actual index. Using these micro-indexes SigLens can search/filter without uncompressing 98% of the data.3) Agile Aggregation Trees.These trees help answer aggregation queries at lightening speeds. These trees are built dynamically during ingestion.

Led the Observability architecture at Salesforce. Scaled the metrics platform from 60 Million datapoints/min to 700 Million datapoints/min. Increased dimension/tags support from 1 Billion to 50+ Billion. Designed various components that were successfully implemented by engineers. Mentored and helped in growing career of 15+ engineers. Presented at various forums.Received President's award for being a top performer in the company (given to top 1%)

Architecting and building the Next Generation Big Data solution for Salesforce logs (Petabyte Scale)

Architect a solution for Enterprise Cyber security

* Designed and built data pipeline that could handle billions of events per day* Implemented various statistics and monitoring alerts* Implemented aggregation analytics on Apache Spark that reduced load on real-time ElasticSearch queries* Implemented various open source tools to interface with Kafka, Elasticsearch, Redis, Couchbase.* Responsible for DevOps of the various data pipeline machines/clusters.

Architected the on-premise version of DataRobot AI platform

* Architected and implemented a distributed multi-petabyte storage solution.* Added a highly available, distributed configuration management system.* Implemented a RESTful interface for managing the object store.* Implemented WebDAV ingest interface* Added Digest and Basic authentication for ingest and delivery interface* Added Microsoft Smooth Streaming, Adobe HDS, Apple HLS streaming protocol support for live TV* Performed stress, redundancy and functional testing* Implemented test infrastructure for continuous integrationVideo Server Product * Increased control plane http delivery performance of the system by 1500 percent* Increased performance of open source web server, YAWS (static content) by approx 80%* Implemented silverlight mp4 parsing algorithm* Responsible for maintenance of the video delivery platform

* Re-Architected the Tatara Convergence Server (TCS) to achieve a 400% increase in performance* Implemented an infrastructure so that complex 3GPP2 call features like call waiting, 3WayCalling, call transfer, call hold for dumb femto cells could be implemented.* Implemented an abstraction layer above the open source sip stack osip* Implemented a memory debug utility to detect memory leaks while the system was under load.* Implemented high availability feature for the call control module.

* Designed and implemented a redundant multi location SIP Registrar.* Designed and implemented a SIP Authentication Server* Designed and implemented an IVR Client/Server* Designed and implemented the Session Border Control (SBC) functionality* Implemented the RSA SecurID two factor authentication service for SIP user agents.* Designed and implemented the TLS/SRTP protocol* Implemented a SIP and RTP Tunneling Server to traverse through highly restrictive firewalls* Implemented call features like CF, CB, CH, CID, TLS enforcement and Call Transfer* Implemented Call Audit feature between various components of SessionSuite

* Integrated next generation SIP features in the Cedarpoint voice communication class 5 switch* Designed and developed the SIP interface to integrate with Application feature servers.* Designed and implemented the call control functionality for features like custom ringback tone,click to Dial, etc. for SIP originated telephone calls.

* Worked on SIP and its associated protocols (like session-timer, sdp, 100rel and the sip info method ) for the SnowShore Voice Media Server.* Designed and implemented the specifications for the SIP and SDP protocol* Designed and implemented the Coach/Whisper conferencing feature* Helped in creating the standard for MSCML protocol (RFC 5022)* Wrote Driver Design specifications for the codec and conferencing features* Wrote drivers for DSP real time voice packet processing.* Debugged and fixed various system level and protocol level problems* Developed several test and diagnostic tools used by QA and development teams