May 2022 – May 2025 · 3 yrs · Bengaluru, Karnataka, India · Remote

• 📊 𝐄𝐦𝐛𝐞𝐝𝐝𝐢𝐧𝐠-𝐁𝐚𝐬𝐞𝐝 𝐑𝐞𝐭𝐫𝐢𝐞𝐯𝐚𝐥 𝐏𝐥𝐚𝐭𝐟𝐨𝐫𝐦: Spearheaded the development of multi-tenant, and multi-entity candidate generation framework. This next-generation platform leveraged the power of embedding-based retrieval to deliver highly relevant results. My role involved:
• 📌 Collaborating extensively with cross-functional teams including Relevance, Infrastructure, and Customer Support to ensure seamless integration and alignment.
• 📌 Led PoCs exploring novel uses of transformer-based LLM embeddings for semantic search and candidate ranking improvements.
• 📌 Specifically evaluating the scalability and performance of the HNSW (Hierarchical Navigable Small World) algorithm, providing actionable insights that directly influenced strategic decision-making for the platform's architecture.
• 📊 𝐆𝐮𝐚𝐫𝐝𝐫𝐚𝐢𝐥𝐬 𝐚𝐤𝐚 𝐕𝐞𝐫𝐢𝐟𝐢𝐜𝐚𝐭𝐢𝐨𝐧 𝐏𝐥𝐚𝐭𝐟𝐨𝐫𝐦
• 📌 Spearheaded the vision and development of Verification Platform designed to ensure robust regulatory compliance and product integrity for launches across diverse global regions.
• 📌 Engineered the system to proactively identify, mitigate, and validate complex regulatory requirements before and during product deployment.
• 📌 Architected and built a dynamic, rule-based engine as the core of the platform, which managed and executed a vast set of configurable rules to verify intricate regulatory configurations specific to each regulatory region.
• 📊 𝐓𝐞𝐜𝐡 𝐄𝐱𝐜𝐞𝐥𝐥𝐞𝐧𝐜𝐞 𝐈𝐧𝐢𝐭𝐢𝐚𝐭𝐢𝐯𝐞𝐬
• 📌 Spearheaded "Food on Nerds" program at India site level, an internal knowledge-sharing forum where engineers presented on diverse technical topics.
• 📌 Curating and hosting engaging fireside chats with tech leaders and subject matter experts, providing valuable insights and fostering open dialogue on emerging technologies and industry trends.
• 📌 Initiating and driving strategic tech initiatives aimed at significantly enhancing code quality, architectural robustness, and system scalability.

Dec 2018 – May 2022 · 3 yrs 5 mos · Bengaluru, Karnataka, India

• 📊 𝐅𝐞𝐞𝐝 𝐈𝐧𝐟𝐫𝐚𝐬𝐭𝐫𝐮𝐜𝐭𝐮𝐫𝐞 & 𝐌𝐋𝐎𝐩𝐬 (𝟑𝟎𝟎𝐌+ 𝐌𝐀𝐔𝐬): Led the design, roadmap, and team execution for real-time feed infrastructure and ML platforms powering ShareChat's core user experience. Delivered the foundational ML platform for personalized feed ranking, enabling scalable model training, fast inference, and consistent user experiences for 300M+ monthly active users.
• 📌 Designed and scaled dual feature stores—low-latency online store for real-time inference and high-throughput offline store for training—ensuring training-serving consistency and rapid experimentation.
• 📌 Architected real-time feed serving systems with sub-100ms p99 latency and high availability, supporting trillions of daily content impressions.
• 📌 Built scalable ranker training infrastructure, including support for deep learning models (e.g., two-tower), powering retrieval and ranking pipelines.
• 📌 Led development of real-time feature freshness pipelines, transforming live user signals into low-latency features to improve model responsiveness.
• 📌 Directed a 30+ cross-functional team across ML, backend, frontend, data engineering, and analytics to ship end-to-end production infrastructure.
• 📊 𝐅𝐞𝐞𝐝 𝐑𝐚𝐧𝐤𝐢𝐧𝐠 & 𝐑𝐞𝐜𝐨𝐦𝐦𝐞𝐧𝐝𝐚𝐭𝐢𝐨𝐧 (𝟑𝟎𝟎𝐌+ 𝐌𝐀𝐔): Led the roadmap & execution for ShareChat’s RecSys stack—spanning MLOps automation, dynamic ranking, and multi-objective optimisation—to drive scalable, high-quality recommendations across 300M+ monthly active users.
• 📌 Led ranker team (e.g., two-tower models) for large-scale personalized feed ranking, with infrastructure for offline training, hyperparameter tuning, and real-time inference.
• 📌 Led multi-objective optimization frameworks using multi-armed bandit-based policy tuning, balancing trade-offs between engagement, diversity, session length, and creator equity.
• 📌 Built CI/CD pipelines for model training, validation & deployment; enabled auto-retraining with drift detection.

Apr 2016 – Dec 2018 · 2 yrs 8 mos · Noida

• Spearheaded a cross-functional engineering team to deliver a 𝐡𝐢𝐠𝐡-𝐩𝐞𝐫𝐟𝐨𝐫𝐦𝐚𝐧𝐜𝐞, 𝐦𝐮𝐥𝐭𝐢-𝐭𝐞𝐧𝐚𝐧𝐭 𝐂𝐌𝐒. This platform became the backbone for India’s leading news publishers at Times Internet.
• Key responsibilities, architectural highlights, and impact included:
• 📌 Powering 𝐦𝐮𝐥𝐭𝐢𝐥𝐢𝐧𝐠𝐮𝐚𝐥 𝐜𝐨𝐧𝐭𝐞𝐧𝐭 𝐨𝐩𝐞𝐫𝐚𝐭𝐢𝐨𝐧𝐬 within an ecosystem reaching over 550 million monthly active users and serving 300+ publications.
• 📌 Mentoring junior engineers and fostering a strong, collaborative engineering culture.
• 📌 Employing 𝐨𝐩𝐭𝐢𝐦𝐢𝐬𝐭𝐢𝐜 𝐜𝐨𝐧𝐜𝐮𝐫𝐫𝐞𝐧𝐜𝐲 𝐜𝐨𝐧𝐭𝐫𝐨𝐥 combined with 𝐝𝐢𝐬𝐭𝐫𝐢𝐛𝐮𝐭𝐞𝐝 𝐭𝐫𝐚𝐧𝐬𝐚𝐜𝐭𝐢𝐨𝐧𝐬 to maintain consistency during high-contention collaborative editing.
• 📌 Implementing a property graph model using 𝐉𝐚𝐧𝐮𝐬𝐆𝐫𝐚𝐩𝐡 to manage complex content relationships, enabling flexible, publisher-specific taxonomies and multilingual tagging at scale.
• 📌 Driving the early adoption of 𝐀𝐩𝐚𝐜𝐡𝐞 𝐊𝐚𝐟𝐤𝐚 to facilitate an event-driven, loosely coupled architecture, supporting near real-time API delivery across web and mobile platforms and improving throughput and fault tolerance.
• 📌 Designing the system architecture to balance consistency, availability, and partition tolerance, optimising for editorial workflow responsiveness and data integrity across distributed teams.

Aug 2014 – Mar 2016 · 1 yr 7 mos · Noida, Uttar Pradesh, India

• During my initial ~2 years in the industry, I immersed myself in designing and building solutions for the TUI group. This period was pivotal for my rapid learning and understanding of complex project intricacies.
• Key contributions included:
• 📌 Dynamic Pricing: Developed a dynamic pricing engine for diverse 𝐭𝐫𝐚𝐯𝐞𝐥 𝐩𝐫𝐨𝐝𝐮𝐜𝐭𝐬 across 4 countries, analyzing complex factors (search behavior, seasonality, popularity) for optimal pricing strategies.
• 📌 Core: Engineered key features (virtual cart, product listings, checkout systems) for 𝐦𝐮𝐥𝐭𝐢-𝐜𝐨𝐮𝐧𝐭𝐫𝐲 operations, ensuring a seamless user experience.
• 📌 Internationalization: Built a scalable 𝐢𝐧𝐭𝐞𝐫𝐧𝐚𝐭𝐢𝐨𝐧𝐚𝐥𝐢𝐳𝐚𝐭𝐢𝐨𝐧 framework enabling multi-language/region support without backend engineering changes, vital for TUI's global operations and efficiency.