Jan 2023 – Nov 2023 · 10 mos · Bangalore · Hybrid

Jun 2019 – Dec 2022 · 3 yrs 6 mos

• Worked as Part of Jetlore - AI Predictive Startup (acquired by PayPal)

Mar 2018 – Jun 2019 · 1 yr 3 mos · Bangalore

• First Mile Algorithm (Supply Chain):
• ==========================
• Scaling the Modern First Mile Algorithm to generate order plans for 10+ Million Item Stores

Nov 2016 – Mar 2018 · 1 yr 4 mos · Bengaluru, Karnataka, India · On-site

• Big Data/Customer Segmentation:
• =========================
• Worked as part of Big Data Team towards Segmenting the Customers.
• Tech Stack: Apache Spark, Python
• Services(Sales/Care):
• ================
• Developed Eligible Offers services (Billing domain) in QuickBooks Online platform. Resource based RESTful services.
• Stripe (Payment Processor) Integration done in Payments Platform.
• Enhanced Customer Interaction Platform Services for Stripe Merchants.
• Tech Stack: Java, Spring & Hibernate

Jul 2014 – Nov 2016 · 2 yrs 4 mos · Bangalore

• Team: Search
• Layer2 ReRanking for InStock and Serviceable Products on Real Time Signals:
• =========================================================
• Designed & Developed the Layer2 to de-rank the Out of Stock and Non-Serviceable products in Real Time
• This is to reduce when the user clicks the top products in Search/Browse results and ends up seeing 'No sellers deliver in this area' and 'Out of Stock' issues
• External Clients(MAPI/Website) call New Layer2. Layer2 gets multiple sellers of top products from Layer1(Solr). Layer2 reranks the available sellers/listings with in a product and de-ranks the products if all sellers/listings are not serviceable in an area (pincode)
• Impact: Reduced the 'No Sellers' issue from 25% to 7% during BBD 2016
• Popularity Ranking of Products (Non-Personalised):
• ======================================
• Revenue per impression based ranking of products in category browsing over the sales based popularity ranking. Handling 70+million products
• Click-Through Rate & Revenue per Click ratios are calculated offline for products and normalised at category level by Hadoop Map Reduce jobs scheduled daily using Oozie
• The metrics are fed into Solr Indexes periodically and products are ranked in Exploit Bucket
• Impact: Popular products featured in browsing results and have improved the funnel conversion ratios
• Near Real Time Data Ingestion (In-Memory updates):
• =======================================
• Apache Storm based Data Ingestion Pipeline to consume real time updates in search systems
• Apache Redis (Jedis) to index the real time data for consumption by Solr and external systems.
• The updates are pushed into Solr indexes and deprecated the single threaded pull based data import handler (which was not scaling to our volume of product updates millions/hr)
• Impact: Reduced the lag in document updates and brought down Out of Stock products in the search/browse results. All the products under the offers reflected on time.

Oct 2012 – Jun 2014 · 1 yr 8 mos · Hyderabad Area, India

• Group: Windows
• Windows OS: Win 8.x , Win 7 & down versions
• ==================================
• Highlights:
• =========
• Worked on Kernel Mode Driver Win32K.sys (Windows Subsystem)
• Fonts(True Type), GDI, GDI+ (Graphics Rendering Engine) APIs
• Customers include Governmental Organizations & Large Private Corporations (Banks)
• Provided OS Kernel Level Security Patches (MSRC: Microsoft Security Response Center)
• Fixed Security Scenarios: Remote DOS, Elevation of Privileges Challenged by Hackers
• Cases involving Race Conditions (Multi Threads) & Double Fetch (TOCTOU) scenarios
• Handled Very Complicated History of Code Changes across 7 OS platforms ( from Windows 8.1 down to Windows 2000) on both GDR and LDR channels for Security Patch Releases
• Kernel Mode Debugging (Windbg,kd)
• User Base: 1,300,000,000 +

Jan 2010 – Sep 2012 · 2 yrs 8 mos · Mumbai Area, India