Jul 2021 – Apr 2025 · 3 yrs 9 mos · India

• Key Takeaways:
• Full-Stack AI Pipeline: Spanned data collection, model development, deployment, and continuous refinement, leveraging both classical ML and state-of-the-art Generative AI solutions.
• Cost Efficiency & Scalability: By applying semantic similarity and vector-based retrieval, significantly reduced computational costs and improved system speed, even with massive data volumes (1+ billion products).
• Business Impact: Elevated user satisfaction through enhanced product discovery and personalization, leading to higher add-to-cart rates and revenue.
• Innovation & Patents: The novel ranking approach (MMOE) and advanced Gen AI strategies for summarization, classification, and retrieval systems contributed to intellectual property filings.
• Cross-Functional Leadership: Demonstrated proficiency in guiding global teams, unifying data engineering, machine learning, and compliance efforts to deliver impactful e-commerce experiences.
• 1. Leading a Cross-Functional Team
• Teams in India and USA: Managed a distributed group of application developers, data/ML engineers, and data scientists. Oversaw end-to-end project life cycles, from design and architecture to production deployment.
• 2. Semantic Query Similarity
• Objective: Reduce computational overhead by grouping semantically similar queries.
• Outcome: Achieved a very hit ratio on query reuse, cutting Gen AI inference costs significantly and ensuring consistent user experiences.
• 3. Personalization & Product Ranking
• Objective: Prioritize products to drive higher conversions, focusing on the first 10 results to impact ATC.
• Outcome: Achieved superior ranking performance, contributing to increased conversions and overall revenue. Undertook patent filing for the innovative ranking solution.
• Overall, these initiatives revolutionized search, ranking, and product discovery for Walmart leveraging advanced NLP, generative models, large-scale embeddings, and robust personalization techniques.

Jan 2019 – Jul 2021 · 2 yrs 6 mos

Jan 2016 – Dec 2018 · 2 yrs 11 mos

• Leading a DSP [Demand Side Platform] is not easy feat but then, that's what keeps me going!! Me and my team breathe and live tech! The undying hunger to learn and grow keeps me humble and on my toes!! We tat InMobi solve problems of high concurrency & massively parallel [QPS upwards of 300K], extreme low latency [all zing bang in 100 ms], predict user value [all things ML/AI]
• We handle 20 Billion of Ad Requests Per day equating to 250,000 queries per second. By the end of this year we may be close to 40 Billion per day which means 500,000 queries per second. We build products that decide which Ad should be shown to which user and what price in less than 100 ms. We process and manage 50TB per day using Big Data Technologies like Spark/Hadoop, Storm, Vertica and whatever it takes.
• We are always looking for smart, entrepreneurial engineers to solve problems using ML/AI first thinking. Join us in our journey at www. inmobicareers.com

Sep 2014 – Dec 2015 · 1 yr 3 mos

• Built Reporting Analytics product - Best product of the year!!
• Working closely with execution leader for Performance product in AdTech that enabled Advertiser to acquire new users. Scale - 200K QPS, 150 ms response time, multi terabytes data logging and processing.
• Built initial version of Video platform for Performance product

Nov 2006 – Sep 2014 · 7 yrs 10 mos · Bangalore

• I work as an architect in User and Data Analytics group in Bangalore.
• Over the last 5+ years I have architected several flagship products like Paid Search BI Stack for MSFT-Yahoo! alliance and Partner Management Center, web portal for Search Partners of Yahoo!, Audience BI Stack [Web Analytics] and STAR dimension modeling on Hadoop+Hive.
• Due to versatility, thought leadership and technical acumen I consistently came up with innovative solutions that not only helped customers to use data more effectively but also provided significant cost benefit to Yahoo!.
• Proposed data usage models blending the strengths of multiple platforms like Oracle, Microstrategy, Hadoop, Hive, HBase and custom tools to address diverse customer needs.
• Instead of building monolithic platforms on expensive Oracle based platform I came up with a paradigm of hybrid solution using Oracle and Hadoop, HBase, Hive that provides infinitely scalable platform at significantly lower cost without compromising on any functionality.
• I work with architects globally to decide on architecture of Digitals Insights at Yahoo (DIY). I was influential in making decision on data access layer which provides flexible, efficient and secure query interface. I also took ownership of architecting infrastructure components like hosted scheduler and negotiated key decisions on other components. This demonstrates my solid technical depth and ability to provide solutions which will be part of the UDA analytical products going forward.
• One of the few architects who has a command on domain [Paid Search Advertiser & Publisher Analytics, Web Analytics] supported by a history of delivering superior results.

Sep 2005 – Nov 2006 · 1 yr 2 mos

• Technology:
• Microsoft .Net 2.x, 3.x , ASP.Net, C#, C++
• SalesForce API
• IIS Server, SQL Server
• Opportunities:
• Build web collaboration application and web marketing with teams in multiple geographic location
• Enhance web CRM application
• Salesforce Integration of Talisma CRM
• Initiatives:
• Drive project plan, architectural design, application design and execution with team in multiple geographic location (India & NewYork)
• Present engineering projects to senior executives and secure the funding for planning and execution
• Build teams, handle project transitions and execution
• Actions:
• Built a web application product – Talisma Chat & Email
• Handle transition project – Knowledge base
• Integrated Talisma CRM using Salesforce API
• Results:
• Successfully delivered web application within timelines and resource constraints

Jan 2004 – Jan 2005 · 1 yr · Bangalore

• Lead team of 10+ DBAs and Application Developers to managed Database and Applications of the following clients:
• Kaplan
• Playboy
• KRD - Knight Rider Digital
• Indepth Knowledge of Oracle Database RAC, Stand By and other Oracle based product solutions.

Mar 2003 – Aug 2004 · 1 yr 5 mos · California, USA

• Lead team of Database Application Developer.
• Handled Tivo device Subscription Database [ OLPT + DW ]
• Involved in feature development, performance optimization, report aggregation and report development for analytical needs.

Mar 2002 – Mar 2003 · 1 yr · Los Angeles, USA

• Worked as Oracle DBA, Performance tuning, optimization, Oracle Apps management

Jan 1999 – Mar 2002 · 3 yrs 2 mos · USA

• Consultant, USA
• Clients: Lucent, Sony, Toyota
• Industry: Telecom, Web based Music [Pressplay], Auto Industry

Sep 1996 – Jan 1999 · 2 yrs 4 mos · Delhi, India

• Sr. Application Developer