A Software engineer focused on building clean, reliable systems that solve real problems. I value long-term impact, fast learning, and working with thoughtful teams.

Programming Languages: Python, JavaScript, R, Bash/Shell scripting, C++

Frameworks: Django, Django Rest Framework, Flask, FastAPI, Tornado, Node.js, Vue.js, Express.js, Electron.js

Database Management: MySQL, OracleDB, MongoDB, ElasticSearch, Redis

Cloud Services and Infrastructure: Amazon Web Services (AWS), Apache Airflow, Nginx, Apache HTTP server

Project Management and Collaboration: Agile Project Management, SCRUM, Git, Technical Recruiting

Other Tools/Technologies: SDXL, Automatic1111, sBERT, FAISS, Celery, R Studio, Power BI, Docker, Plumber API Generator, PM2

Engineering Manager at V-Mart Retail Ltd. (2022-)

Product Information Management (PIM) Platform

• Designed and implemented a robust ETL pipeline using Python and Airflow, exposing services via FastAPI and persisting data in MongoDB to ingest product images from multiple sources/stages — purchase orders, warehouse GRC/Inwards, and photoshoots — ensuring strict adherence to image SOPs at every touchpoint.
• Leveraged AI models like Google’s Gemini image-to-text to extract rich product attributes from images, enabling automated tagging, accurate classification, and enhanced catalog quality.
• Built a bi-directional catalog normalization layer to harmonize third-party marketplace data (Myntra, AJIO, etc.) with LimeRoad’s internal taxonomy, and to export LimeRoad’s catalog in marketplace-compatible formats.
• Led the development of a product similarity engine using sBERT embeddings and FAISS-based vector search, powering recommendation systems and optimizing ad targeting through intelligent product grouping.
• Implemented semantic search with vector-based queries for enhanced product discovery.
• Automated image enhancement workflows — including upscaling, background generation, and retouching — improving product visuals across SEO, marketing, and Product Description Pages (PDP).
• Developed an AI-assisted mobile notification banner generator, using Computer Vision (CV) and Stable Diffusion (SDXL model) to compose creatives aligned with product aesthetics.
• Programmatically enriched product titles, descriptions, and attributes, driving better SEO performance and conversion rates on product detail pages.

       Tech Stack: Python, FastAPI, MongoDB, Airflow, sBERT, FAISS, Gemini, OpenCV, SDXL

Event-Driven Customer Journeys Platform

• Architected and led the development of a real-time, event-driven customer engagement system that powers dynamic journeys such as birthdays, referrals, anniversaries, high bill amounts, etc — directly contributing ₹50Cr+ per year in revenue through automated, personalized touchpoints.
• Designed a modular microservices architecture using FastAPI and Kafka to decouple event producers and consumers, enabling high throughput and horizontal scalability across multiple customer touchpoints.
• Implemented a JSON-driven Rules Engine using MongoDB to store and evaluate event-action mappings, allowing non-engineering teams to define new journey logic without code changes.
• Built a dynamic Event State Management layer (MongoDB-based) to track context-specific states like referral completions, coupon usage, and action retries — critical for accurate and idempotent execution.
• Developed extensible Action Resolvers as FastAPI-based microservices that consume events from Kafka, evaluate rules and invoke downstream systems (e.g., coupon engine, communication platforms) in real time.
• Integrated communication workflows via Karix (SMS/WhatsApp) and LimeRoad’s mobile app, orchestrating millions of high-conversion, contextual messages at scale.
• Leveraged our OLAP DB ClickHouse for real-time event logging and analytics, enabling granular visibility into system health, conversion performance, and customer behavior across journeys.
• Designed the platform to be highly extensible — supporting easy onboarding of new event types and engagement actions with minimal engineering overhead.

        Tech Stack: Python, FastAPI, Kafka, MongoDB, ClickHouse, REST APIs

Coupon & Offer Engine

Architected and scaled a comprehensive promotion management system powering ₹200 Cr+ in annual revenue, enabling seamless creation, configuration, and delivery of coupons and direct product-level discounts across omnichannel retail.

• Built a robust engine to support coupon-based and non-coupon (flat & relative) offers such as Buy 1 Get 1, Flat ₹400 off, or 20% off, with support for percentage, fixed-amount, and free-product reward types.
• Modeled offers using a normalized mathematical framework (x → condition, y → reward) for accurate, extensible pricing logic — inspired by scalable commerce engine design.
• Developed a metadata-driven, table-per-assortment architecture with materialized/union views to support ultra-fast offer resolution across thousands of active promotions.
• Enabled multi-level coupon issuance with dynamic eligibility logic and store-specific campaigns, allowing granular targeting and layered promotional strategies.
• Integrated real-time offer and coupon querying via WhatsApp, SMS, and missed-call flows, greatly improving customer accessibility and campaign visibility.
• Provided an authenticated coupon search interface for store staff with tracking and resend capabilities to improve customer service and accountability.
• Automated store-level data pipelines from the central data warehouse, ensuring synchronized targeting and real-time campaign availability across all retail touchpoints.

• Deployed and maintained production systems on AWS EC2 with full ownership of infrastructure, deployment pipelines, and logging, ensuring high availability.

  Tech Stack: Node.js, Express.js, MongoDB, ClickHouse, Airflow, AWS EC2, Linux, REST APIs

ClickHouse-Powered Central Analytics Platform

• Designed and implemented a centralized data lake with ClickHouse as the OLAP engine to support the organization’s cross-functional analytics and decision-making at scale.
• Built real-time ETL pipelines to ingest data from multiple source systems — including sales, inventory, products, stores, and customer activity — into a unified ClickHouse-based lakehouse architecture.
• Using granular sales patterns, enabled high-performance analytics for critical business functions such as store-level assortment planning, inventory optimization, and product performance.
• Powered revenue tracking via coupons, purchase behavior analytics, and customer segmentation, establishing ClickHouse as the analytical source of truth.
• Delivered real-time customer recommendations on LimeRoad using behavioral data streamed into ClickHouse, improving user engagement and conversion.
• Created “till-date” and “rolling last-year” RFM (Recency-Frequency-Monetary) profiles for 50M+ customers, enabling advanced lifecycle marketing, retention, and personalization.

• Designed ClickHouse schema for high-cardinality, high-volume workloads, ensuring sub-second analytical queries across billions of rows.

  Tech Stack: ClickHouse, Python, Airflow, MongoDB, REST APIs, ETL Frameworks, Retail POS Integration

HR Schemes & Incentive Manager

• Designed and led the development of an end-to-end incentive management platform for 500+ retail stores, automating incentive distribution across thousands of store employees based on real-time sales data from the POS system.
• Engineered dynamic scheme configurations that allow business users to assign specific products and incentive logic by employee role and designation, ensuring accurate, personalized incentive computation.
• Built a real-time sales and incentive tracking engine by integrating directly with retail POS systems, enabling instant visibility into target vs achievement metrics across employees and stores.
• Implemented predictive analytics models to forecast store-level sales performance during festive periods using historical data, improving resource allocation and planning accuracy.
• Developed a configurable communication layer to dispatch updates (e.g., incentive progress, sales targets) via SMS and enterprise chat (PeopleStrong Jinie) with flexible scheduling (e.g., every 3 hours for 2 months).
• Enabled role-based access control with granular permission layers — from store managers to department heads — replicating enterprise-level maker-checker workflows and ensuring data security.
• Delivered a user-friendly interface empowering non-technical business users to create, launch, and monitor incentive schemes, reducing turnaround time from days to hours.

• Provided real-time dashboards for performance tracking and reporting across stores and individuals, enabling data-driven decisions and faster incentive cycles.

  Tech Stack: JavaScript, Express.JS, MongoDB, REST APIs, POS integration, PM2

Distributed Data Lake Gateway (DDLG)

Led the conceptualization, design, and supervision of the development of a native application facilitating seamless exchange of files and processes across all organizational nodes.

• Implemented global search and file-sharing functionalities, enabling efficient retrieval and collaboration on files. 
• Implemented user access management, ensuring secure and controlled access to the application's features.
• Defined the structure for state management of pipelines and Data Transformers (DT), optimizing data processing and management.

• Diligently supervised the project, ensuring deliverables at every stage of the development journey.

  Tech Stack: Python, FastAPI, MongoDB, Electron.js, React.js, Redux

Store Barcode Utilities

• Built and deployed a suite of barcode and QR code scanning apps to streamline store operations such as returns, shipments (AWB), and product tracking.
• Developed in Streamlit with real-time scanning support and a fallback OCR mode to extract data from images when scanning fails.
• Synced all scanned data to a centralized backend, enabling the reconciliation team to track scanned items store-wise — eliminating the need to manually verify claims with store staff.

• Reduced reconciliation time and manual errors, increasing accuracy and saving hours of effort for both store and backend teams.

  Tech Stack: Streamlit, OpenCV, Tesseract OCR, Python, MongoDB

AI-Powered File Interaction & Content Generation

• Built an internal tool for design, category, and content teams to interact with product images and data files for generating titles, descriptions, and copywriting using AI.
• Enabled chat-based interaction with image and metadata files, streamlining collaboration across creative and merchandising workflows.
• Integrated text-to-image models (SDXL, AWS Titan) for use cases like creative ideation and visual generation.

• Deployed a custom GUI for SDXL called Automatic1111 on an on-prem machine, allowing users to generate and edit visual content without technical intervention.

  Tech Stack: Python, Streamlit, Stable Diffusion XL (SDXL), AWS Titan

Happiness Agent Dashboard

• Led the development of a dynamic web application, utilizing FastAPI, to track the real-time performance of store staff in the "Happiness Agent" project. This initiative aimed to boost online sales by engaging store employees in an endless aisle program, encouraging them to promote and entice customers to order from the brand.com website.
• Designed and implemented a user-friendly dashboard providing real-time visibility into the performance of store staff, enabling corporate offices to make informed decisions and take appropriate actions.
• Developed a robust ETL pipeline for online sales data collection and extraction of Happiness Agent orders. Presented region-specific insights, enabling informed actions by regional managers.

• Designed an intuitive UI for the dashboard, enabling a good user experience. 

  Tech Stack: Python, FastAPI, MySQL, Bulma CSS, Jinja Templating

Full-Stack Software Engineer at Lyxel & Flamingo (Jun 2018 – Jan 2022)

• Built Lyrkl, a social media scheduling and analytics tool with Python Tornado and Vue.js.
• Added Celery-based scheduling, sentiment analysis, and brand monitoring.
• Built Instagram influencer search engine.
• Unified analytics via integrations with Facebook, Instagram, Twitter, and LinkedIn.
• Secured using OAuth2; used hybrid SQL/Mongo architecture.