Sharad Jain

Mentoring students in AI and machine learning, providing guidance on building production-ready AI systems, agentic workflows, and advanced AI techniques.

Architected and deployed a production-grade, agentic conversational AI for real-time voice, achieving <500ms latency by orchestrating LiveKit, Twilio, and Google's streaming Gemini API. Engineered advanced RAG system to ground text-to-SQL agent in complex database schemas. Designed structured communication backbone for multi-agent workflows with OpenAI function-calling. Developed automated multimodal content pipeline using visual AI (GPT-4.1-mini) for frame-by-frame video analysis, reducing manual content creation efforts by over 80%.

Architected data-driven growth infrastructure for wedding platform serving 500K+ couples, implementing cohort-based retention models that increased 30-day user retention by 45%. Built experimentation framework with statistical rigor, running 15+ A/B tests quarterly that improved conversion funnel by 28% and reduced CAC by $120. Designed behavioral analytics system using Mixpanel and SQL to identify activation moments, resulting in 2.3x improvement in user onboarding completion rates. Implemented product-led growth loops connecting vendor engagement to organic referrals, driving 35% of new user acquisition through viral mechanisms. Spearheaded cross-platform marketing attribution, achieving 95% accuracy in conversion reporting and 20% YoY channel performance improvement.

Implemented MLOps practices for enterprise engineering team, including automated CI/CD pipelines, resulting in 40% faster model deployment and 30% reduction in production issues. Engineered scalable data pipelines using Apache Spark and Airflow, processing 10TB+ daily data, improving data freshness by 50% and model accuracy by 15%. Optimized user onboarding through A/B testing, reducing activation time from 7 days to 2 days and improving retention by 30%.

Implemented machine learning models to predict transient infrastructure failures. Improved prediction accuracy by 15% through feature engineering and model optimization. Built AWS Quicksight dashboard for business insights and interfaced with Data Platform Team in predictive maintenance.

Designed and implemented complex ETL mappings using IBM Infosphere DataStage for large-scale data warehousing projects. Performed detailed impact analysis on existing ETL mappings in response to evolving business requirements. Mastered data extraction techniques across multiple data repositories, significantly reducing data retrieval times for analytical reporting.