Sep 2021 – Mar 2025 · 3 yrs 6 mos · Bengaluru, Karnataka, India

• SymphonyAI operates across multiple verticals (Retail, Media, Finance, Healthcare, Manufacturing, ITSM) with a mission to create high-value enterprise AI solutions with fast time to value. Accelerate is the horizontal organization within SymphonyAI that enables this journey by building the necessary platforms and technologies.
• Accelerate-Eureka is a cloud-neutral, multi-tenanted platform that enables building and operating AI-driven enterprise applications at scale. Its core capabilities include DataOps & MLOps (data pipelines, model deployment, monitoring etc), Auto-ML (automatic model tuning), Observability (drifts, trends etc), Multi-Modal Predictions (batch, stream or on-demand), built-in Feature Store and LLM/Generative-AI driven solutions. Eureka also provides an integrated ecosystem to experiment with data and build new algorithms/pipelines, a low-code application building framework and API/SDK based integrability.
• As the chief architect for Accelerate, I have been primarily responsible for design & architecture of the new generation Eureka Platform. My role also includes advanced technology research, supporting new product incubations, guiding and mentoring junior architects and providing technology guidance to vertical focused organisations.
• This role brings in deep hands-on exposure to both established and cutting edge technologies like GenAI, LLM, MLOps, Iceberg, Trino, Minio, Kafka, Kubernetes, Kubeflow, MLFlow, Dask, Feast, Spark, Seldon and multiple advanced frameworks in machine learning, deep learning

Sep 2018 – Sep 2021 · 3 yrs · Bengaluru Area, India

• Clustr is a part of Tally Solutions, focused on bringing the power of Big Data and Artificial Intelligence to small businesses. At Clustr, we crunch humongous amounts of data and apply cutting edge deep learning techniques to achieve things way beyond the current standards of our industry
• My responsibilities include overall architecture and strategy of data science and engineering. these range from defining abstract design principles all the way to building POCs and and guiding engineering teams to implement our vision

Oct 2016 – Aug 2018 · 1 yr 10 mos · Bengaluru Area, India

• Legilimens is a semantic search and discovery engine for online commerce. It outperforms on-site searches of most merchants and is a potential game changer for mobile commerce.
• Traditional search is constrained by natural limitations of text based indices - even with synonym augmentation. As a result, for most online commerce sites, search quality drops drastically as queries get somewhat complex. Not only does this hurt the top line, it is also a big dampener for voice based searches and m-commerce. This is Legilimens comes in.
• Legilimens is based on multiple patent-pending breakthroughs in artificial intelligence(AI) and natural language processing(NLP). Our human-machine collaborative learning framework consumes huge volumes of diverse public data to learn a deep, multi-dimensional domain vocabulary. Our probabilistic semantic indexing and intent mapping technologies helps interpret context & intent of queries, as well as identify distinctive attributes of merchandise. Overall result is a significant improvement in accuracy and recall even for vague, ambiguous searches.
• Legilimens Search Engine has been validated with real data. Proof of concepts, based on data crawled from Myntra and Tesco, outperforms both the merchants' on-site searches by a fair margin.
• Article on Legilimens : https://www.linkedin.com/pulse/draft/AgGDUs5tSt3UugAAAXE1nbzvhl7gmXSZ3UPDq8fqfg7T3CpPyWYAAFlLjYerxkEJuZqc0so

Sep 2014 – Aug 2015 · 11 mos · Bangalore

• DevFactory is a pioneer in remote workforce model, where a globally distributed virtual team comes together, either for a short term assignment or longer term product development.
• I was part of the team that built the tracking software for the Crossover talent management platform. The purpose of this software is to track and measure actual productivity of remotely located contractors so that fair billing is ensured. As a chief architect, my role was to provide technology leadership to a globally distributed team as well as take hands-on role significantly. Technologies involved were Java, AWS and machine learning.

Oct 2013 – Aug 2016 · 2 yrs 10 mos · Bengaluru Area, India

• Ruaha is an early stage distributed ML Ops Platform, built on Hadoop. It provided an end-to-end ecosystem to integrate data, train models and deploy large scale machine-learnt solutions. With a lot of standard machine learning functionality built in, Ruaha could helps break the sandbox-to-production barrier and enables enterprises leverage data science at scale. Through its comprehensive abstraction framework, it makes it possible to plug-in third-party algorithms, thereby helping adopt technology advancements rapidly. Further flexibility is achieved through custom rules, pre-processors and compound models.
• Capillary Technologies is a major provider for customer engagement software and services in the retail space. Ruaha Platform was acquired (acqui-hire) by Capillary in September 2015 to enable intelligent campaign targeting for their customers.
• Technologies/Platforms/Software: Core Java, J2EE, Hadoop, AWS[EC2, S3, DynamoDB, SQS, EMR] MySQL, Quartz, Storm, Mahout, Redis, Vadin, Random Forests, Collaborative Filtering.

Jan 2013 – Sep 2013 · 8 mos · Bengaluru Area, India

• Retail Pricing Systems
• Pricing platforms support Amazon retail worldwide and employ versatile pricing strategies - from simple markups to competitor matching and adaptive markdowns. Challenges include extreme scalability and reliability requirements, as well as stringent accuracy needs.
• Technologies/Platforms/Software: Java, J2EE, AWS [EMR, Dynamo DB, S3, SQS], MySQL, Random Forest

Apr 2010 – Nov 2012 · 2 yrs 7 mos · Bengaluru Area, India

• Merchant Services Dashboard
• Multi-tenanted platform to enable Yahoo business subscribers [E-Commerce, Hosting Services, Listings] to acquire, integrate, track and analyze data from a host of partner services. Challenges include the massive scalability requirement to support about 15 million merchants, heterogeneity of integration points and reliable management/tracking of service life cycle.
• Technologies/Platforms/Software: Core Java, J2EE, Hadoop, Sherpa, MySQL, Quartz, SCXML

Jan 2008 – Jan 2010 · 2 yrs

• Category Analytics
• Category Analytics Platform provides statistical analysis, prediction and report generation in diverse areas like trade promotion, stock outage, price-volume elasticity and customer sentiments specific to Retail and CPG domain.
• Technologies/Platforms/Software: Core Java, J2EE, Oracle, R Statistical Software, Statistical Models and Algorithms, MSBI Analytics Platform

Jan 2007 – Jan 2008 · 1 yr · Bengaluru, Karnataka, India

• Design and Architect, enterprise security solutions in the areas of Identity & Access Management, Enterprise Security Architecture, Security Information & Event Management, Data Privacy, PKI, Encryption etc.

Jan 2004 – Jan 2007 · 3 yrs · Bengaluru Area, India

• Compute Pool Manager
• Azul Systems was a global provider of enterprise scale multi-core appliances that deliver processing and memory resources as a shared network service to transaction-intensive applications. Built on the Java platform, it enabled transparent, massively scalable infrastructure. The key enabler for network attached processing is called the Compute Pool Manager (CPM) – a piece of replicated, highly available software that handles provisioning and monitoring of the appliances and provides a virtualized interface to client applications.
• Technologies/Platforms: Core Java, JCA, JDBC, XML, HTTP, SSL, PostgreSQL, Tomcat, Servlets, JSP, Struts, RMI etc.

Jan 2001 – Jan 2004 · 3 yrs · San Francisco Bay Area

• Firefox Browser and AOL Communicator Security
• Support for key and certificate management [PKI, PKCS], SSL and S/MIME for Firefox browser and AOL Communicator
• Technologies/Platforms: C++, wxWindows [a cross-platform UI framework], NSS, AOL’s proprietary client side libraries