Akshay

Staff-level Applied Scientist with 9+ years of experience architecting & leading production machine learning and LLM systems at enterprise scale across SaaS, marketing, sales, and airline domains. Open source contributor to leading AI projects with experience partnering with senior stakeholders & translating first-principles research into business-critical AI systems through rigorous experimentation and explainable modeling.

Senior Data Scientist at 6sense (2024-09 – Present)

• Agentic AI Email Agents & Autonomous Response System: Architected and owned the strategic design of production-grade agentic LLM systems with intent detection and response generation and comprehensive guardrails, achieving 96% QA pass rate through rigorous evaluation frameworks
• Reduced manual intervention by 60% and lead response latency by 2×, enabling faster SDR follow-up and improving reply SLAs through rigorous engineering and evaluation
• Tech Stack: Python, FastAPI, Docker, LLMs, Agents, GitHub Actions
• Multilingual Job Level Classification System: Architected end-to-end DistilBERT-based multilingual job title classifier processing 5M titles/week on AWS SageMaker with sophisticated 7-level job hierarchy
• Improved lead scoring precision by 7%, expanded persona mapping coverage by 15%, and reduced customer issues by 50% through scientifically-grounded model development while mentoring senior data scientists and engineers
• Tech Stack: Python, SageMaker, Transformers, BERT, Optimum

Senior Data Scientist at Freshworks (2022-03 – 2024-09)

• NER & Intent Engine: Architected and led applied NLP research translating multilingual NER models (BERT + dependency parsing) into production systems across software, hardware, and HR domains
• Achieved 84% F1 score (strict match), 30% improvement in search results, and 20% improvement in ticket deflection through systematic, data-driven methodology
• Tech Stack: Python, spaCy, Transformers, LangChain, ChatGPT, FastAPI, BERT

Lead AI & ML Engineer I at Delta Airlines (2020-02 – 2022-03)

• Destination Recommendation System: Architected hybrid recommendation engine using Deep Neural Networks and BERT, deployed on AWS SageMaker with 5M API hits capacity
• Achieved 50%+ top-25 recall and 40% improvement in revenue per email sent through rigorous experimental design
• Tech Stack: Python, TensorFlow, SageMaker, AWS Lambda, API Gateway, Locust.io

Data Scientist at Brillio (2016-02 – 2020-02)

• Sales Planning Optimization: Led cross-functional team in developing inter-hierarchy time series forecasting architecture for inventory optimization
• Reduced customer wait time by 15% and holding costs by 12% through optimized inventory stock predictions
• Tech Stack: R, Time Series Forecasting, hierarchical time series (hts)

Master of Economics in Economics – Delhi School of Economics, University of Delhi (2015)

Bachelor of Economics in Economics – Hansraj College, University of Delhi (2013)